High-quality patient-centered care is widely recognized as a priority in health care [1] and has been shown to improve patient experience, patient safety, and accessibility to services [2,3]. Health care professionals engaged in patient-centered care focus on ensuring that patients’ experiences at a physician, hospital, or rehabilitation facility meet basic standards of care, such as treating them with courtesy, informing them about their care, and minimizing pain during their visit. Standardized questions surrounding these aspects of a medical visit are widely used within the health care industry to inform patient-centered care best practices and to improve governance, public accountability, and patient autonomy [4,5].

Paper-based collection of experience surveys remains time-consuming, expensive, and limited by factors such as nonresponse, recall bias, and inadequate sample size [6-8]. Although paper-based surveys may be advantageous in some circumstances, that is, to reduce startup costs or for nondigital natives such as the elderly, the use of smartphone-based survey apps, which we call smart surveys, provides new opportunities to improve data collection techniques. Mobile technology overcomes many of the limitations of paper-based surveys and enables collection of large quantities of real-time data over a broad geographical area. Exploiting this technology opens the possibility of private and public sector services, health care providers, and government bodies effectively engaging with the public, one-on-one, to better respond to their needs. However, at present, there is little guidance to help service providers understand when and where individuals are willing to disclose service experience data using their smartphones. In particular, there is a lack of understanding of users’ beliefs, perceptions, and attitudes toward sharing of health service experience feedback using their smartphones.

Patient experience surveys are validated questionnaires, which are developed by health services experts to understand patient perceptions of their health care experience and serve an integral role in patient engagement and service improvement [3]. For example, the WatLX patient experience questionnaire [9-11] determines agreement with statements such as “I was always treated with courtesy,” “My physical pain was controlled as well as possible,” or “From now on I know what to expect about my care.” When such patient experience feedback is collected, collated, and interpreted, findings can drive critical and necessary improvements in service quality, patient safety, and clinical effectiveness [3,4]. In recent years, to increase the benefits yielded from such surveys, health service research has begun to shift from “traditional” paper-based methods of survey administration to the use of technology-enabled survey tools. Yet there is little published research on the use of smartphone technology to collect individuals’ experiences of health services.

The health care literature has found no significant differences in data equivalence or validity between paper- and Web-based surveys [12-14]. With mobile devices, researchers are able to collect large quantities of data over broad geographical areas. Furthermore, the use of smartphone-based apps provides access to functionality such as location-based activity detection [15,16] and notifications [17] that can help improve survey compliance and completeness of survey responses to improve the overall reliability of results [18]. In situ assessment of services, where data are collected during or immediately after a service encounter, may yield an even more immediate and insightful level of understanding of service quality [19]. However, the context in which these data are collected can also introduce new issues such as concerns around privacy [20].

Smart surveys, which we define as smartphone survey apps that use advanced functionality to provide more intuitive surveys or gather more contextual complementary data in addition to participant responses, introduce a number of unfamiliar behaviors for patients to undertake that may work as barriers to adoption. For example, they require users to download an app to their personal smartphone and to disclose potentially sensitive information via a digital channel that may be perceived as public or insecure. To better identify and understand how these barriers may impact the adoption of smart surveys, we turn to the theory of perceived risk [21,22].

A user’s perceptions of risk can have a negative effect on information system (IS) adoption [21,23,24]. Initially introduced in the context of consumer behavior research, perceived risk can be conceptualized as the subjective expectation of loss experienced by a consumer during purchase decisions [21]. Perceived risk describes risk as a multidimensional construct that comprises facets such as financial risk, performance risk, physical risk, psychological risk, social risk, time risk, and privacy risk. Our framework of perceived risk is adapted from studies by Jacoby and Kaplan [22] and Featherman and Pavlou [23] (Table 1).

Consumer behavior and IS research has found perceived risk and its antecedents to be key predictors of electronic service adoption; for example, perceived risk and its dimensions are inhibitors of technology acceptance model variables [19]. Of all the facets of perceived risk, privacy (security) risk is demonstrated to be the most important barrier in the adoption of e-services for consumers, having both direct and indirect influences on the intention to adopt [25]. Financial risk, the second most important inhibitor to adoption, also has a significant negative impact on attitude to technology adoption [26]. Time risk has a negative influence on attitudes, implying consumers are concerned about delays and length of time to complete a transaction.

Perceived risk impacts attitudes toward adopting mobile e-services [27-29] as well as the intentions of use among both frequent and infrequent users of mobile e-services. In these studies, results have consistently shown overall perceived risk to be mediated by privacy, financial, time, and performance risks. With respect to mobile health (mHealth) app adoption studies, perceived risk has significant and negative effects on attitudes toward adoption. For example, Schnall et al [30] identified patient concerns regarding security (eg, health information or location sharing) when referring to mHealth apps and smartphone devices, similar to Zhou [31].

Previous research has categorized users based on the intensity of their perceptions of privacy risk. Westin [32] separates technology users into 1 of the 3 risk groups based on their willingness to share personal information on the Web: (1) privacy fundamentalists (high privacy orientation and supports regulatory controls), (2) privacy pragmatists (weigh benefits to self or society and bases trust on context), and (3) privacy unconcerned (willing to share information and reject privacy concerns). These categories were developed over a series of more than 30 consumer surveys [33] and have been used extensively by the computer-human interaction community as a requirements engineering design tool to help anticipate user needs and design functionally relevant technology. However, in practice, this methodology has faced criticism, and research has found a lack of correlation between Westin’s categories and user behaviors and attitudes (ie, willingness to share information), perhaps attributable to the development of the questions before the internet [34-36]. In response to this mismatch between Westin’s categories and user behavior and to more fully account for the importance of privacy risk as a barrier to adoption of smart surveys, we turn to Dupree et al’s work on privacy personas [37,38].

Dupree et al [37] developed privacy personas to add contextual information to participants clustered by their attitudes and behaviors related to security and privacy. The personas provide a better understanding of a user’s proactivity and ability to act upon privacy risk concerns, better aligning their behavior with their attitudes. Dupree et al identify the following 5 clusters:

Morton and Sasse [39], who performed their research concurrently with Dupree et al, also identify 5 clusters that closely correspond to those listed above but in the context of location disclosure.

The purpose of this study was to understand what beliefs, perceptions, and attitudes influenced patients’ intentions to share health service experience feedback using their smartphones, in particular, what role perceived risk plays in this process. Health care providers are increasingly being held accountable for the quality of services they provide; however, data collection is expensive, response rates are low, and turnaround times can be long. Although mHealth apps are common in the sector, and smartphones have been used to collect experience data in other industries [40,41], there has been little research into the use of mobile apps to collect in situ, location-based experience data in health care.

We recruited participants from a local university between January and February 2017 using posters, email, and snowball sampling techniques. Participants were classified according to their privacy persona and their dimensions of perceived risk, and their responses were sequentially analyzed to allow researchers to evaluate the breadth of our sample and to ensure that individuals with different technical backgrounds as well as varying degrees of privacy tolerance related to information sharing were included in the study. Recruitment and analysis proceeded until saturation [42,43]. Participants received Can $10 for participating in the interview.

Participants were welcomed upon their arrival and were given an overview of the purpose of the study and the data collection process by a researcher. Participants reviewed and signed an informed consent form and provided demographic information. Data were collected from participants using short questionnaires followed by a semistructured, in-depth interview. The information gathered from the questionnaires and think-aloud technique provided information for classification of participants into privacy persona clusters as well as complementary data for further context.

In the first questionnaire, participants were asked to rate their perspectives on privacy and security (PAS) (Multimedia Appendix 1). The second questionnaire included Westin’s privacy user questions [32] and questions that were related to their knowledge and motivation to protect their privacy (Multimedia Appendix 2) [37]. Before completing the final questionnaire and interview, participants were given the opportunity to learn more about the mobile app used for the survey distribution, analysis, and administration [44] and to evaluate its key features and user interface. The MetricWire app is a commercially available mobile phone and Web-based survey platform [44]. The app allows mobile completion of surveys, with functionality such as automatic alerts or triggers to prompt phone owners to respond to a survey based on factors such as time of day or location of the device. Researchers loaded a modified standardized and validated patient experience survey [9,10] into the MetricWire platform using a smartphone provided by the researchers (Figure 1). Participants were asked to recall their last visit to a physician and fill out the survey. Completion of the survey ranged from 2 min to 5 min.

Using cognitive interviewing techniques [45] (think aloud), where questions are administered, and participants are encouraged to verbalize the reasoning behind their answers, a third questionnaire (including questions adapted from Jacoby and Kaplan’s [22] perceived risk study), was used to assess participants’ perceptions of risk (Multimedia Appendix 3). These semistructured participant interviews ranged from 20 min to 40 min. Throughout and at the end of the interview, the researcher summarized their interpretation of each participant’s responses. Participants were encouraged to add any additional information that they felt was missing from the summarized interview responses. This process served as an informal method for member checking [46].

Upon completion of each participant interview, participant responses were transcribed manually from the digital recordings and thematically analyzed using QSR International’s NVivo 11 [47]. Responses were used to inform subsequent interviews. We used constant comparison and content analysis to code and analyze the transcripts [48], with 3 researchers (DN, JM, and JW) reviewing the interview transcripts independently and using consensus methods to iteratively discuss content and discrepancies to ensure coding consistency. Data were manually sorted using the perceived risk framework and then thematically analyzed to uncover the unique challenges facing smart surveys’ adoption and use for patient experience sampling. Interview transcripts underwent initial open-ended coding where quotes were divided into 4 concepts and 12 subconcepts based on similarities in meaning or context. These concepts were then discussed among the researchers to further develop emergent themes. The themes were developed based on both the perceived risk typology (deductive reasoning) and open and axial coding of interview transcripts (inductive reasoning) [47,49].

To assign participants to 1 of Dupree’s clusters, each participant’s data were reviewed (DN), and the participant was preliminarily assigned. Following a process of discussion (JM, JW, and PM), participants were reassigned as necessary until each cluster remained stable. Participant data were collected until we had at least one participant from each of the Dupree clusters identified, and saturation was achieved, where no new themes or evidence emerged from the interview transcripts [42,43].

Ethics approval for this study was sought and obtained jointly from the ethics committees at Wilfrid Laurier University and the University of Waterloo (#4690). All participants provided written informed consent before participating in the study.

We conducted 24 semistructured interviews with Canadian smartphone owners (7 male and 17 female) with varying educational backgrounds, technical knowledge, and motivations to protect privacy. All the participants were registered university students, half at the graduate level and, as such, are “digital natives” and thus confident using smartphones and mobile apps [50]. All had received some postsecondary or postgraduate education, were comfortable speaking English, and regularly used a smartphone. The average age of the sample was 23.9 years (minimum: 21 years, maximum: 56 years, and median: 22.5 years).

Respondents were classified according to Dupree et al’s [37] privacy clusters framework (Table 2). Participant 4 was rated with low technical knowledge yet high motivation to protect their privacy. Therefore, they did not fit any of the Dupree privacy personas, and we classified them as “Undefined.”

To develop an understanding of the core issues facing smart survey adoption, we also categorized responses according to the dimensions of perceived risk [21,22] (Table 3). Participants were most concerned about PAS and performance risk; under the circumstances, that is, the introduction of a novel app for use in a contextually sensitive location, this result was predictable. None of the participants expressed concerns related to their psychological or social well-being as a result of using the app.

A number of themes emerged from our analysis of cognitive and in-depth interview transcripts: (1) perceived risks associated with smart survey use, (2) loss of information agency, and (3) trusted data collectors and altruistic intentions. These are organized according to the focus of this study: first, how perceived risk impacts the propensity to use smartphones to provide service feedback using our perceived risk framework, and second, the role of participants’ key beliefs, perceptions, and attitudes in that process.

Mobile apps are increasingly being used to gather real-time clinical and ecological patient data and to help manage workload in the health services sectors [50-54]. Although smartphones are changing the way we deliver health care and engage patients [55], examples of their use in collecting in situ patient service experience data are scant. We explored perceptions and attitudes, which impact the adoption and use of smartphone-based apps to collect patient experience data, referred to as smart surveys. The theory of perceived risk [21,22] suggests there will be inhibitors to smart surveys adoption. Yet, although participants mentioned perceived risks normally associated with electronic commerce and other Web-based activities such as social, psychological, physical, financial, and time risks [23,56], these were considered minimal.

The study used an app that participants would have to download, retain on their phone, and manage alerts and the software app itself over time. The most commonly perceived risk was PAS, consistent with other mHealth and wearables literature [51,57,58]. Participants’ PAS risk perceptions were related to connectivity, data sharing, encryption, and storage. Consistent with Dupree et al [37], we observed that not all digital natives have the same level of technical knowledge. Our thematic analysis revealed other factors indirectly related to risk that influenced participant’s perceptions of smart surveys as a conduit for patient service experience data collection. In particular, participants trusted the data collectors and communication channel, thereby reducing perceptions of risk. Conversely, concerns over loss of information agency, evoked based on past social and personal experiences where they lost control of their data, served to amplify their PAS concerns. These themes have implications for the design and use of smart surveys apps in practice in 3 main areas. The 3 themes, with recommendations for developers are as follows:

The themes identified through our interviews helped to develop an understanding of barriers to smartphone-based patient experience surveys. Nevertheless, we are careful to acknowledge limitations to this study. First, the attitudes and perceptions of risk held by our participants were captured at 1 point in time, and attitudes toward adoption can change over time and as the y become more familiar with technology [63,64]. Furthermore, participants were not asked to download and use the app on their own devices. Consequently, there may have been less consideration given to risk as participants did not actually surrender any personal information. Future work will address these limitations through data triangulation and a longitudinal validation that patient behaviors reflect their reported perceptions. Second, participants were mostly younger (average age: 23.9 years) and more educated, and as a result, they are not necessarily representative of the largest patient segment using health care services [65]. Although the results of this study may not be generalizable to the wider population, a majority of adults now own cell phones, 77% of them own smartphones, and a growing percentage of adults aged 50 to 64 years and over 65 years are smartphone users (73% and 46%, respectively) [66]. These rapidly growing rates and the decreasing availability of cell phones without expanded digital capability and access to the internet suggest that adults across the life stage will soon experience similar issues as the study sample.

We used nonprobability convenience sampling and a nonsystematic recruitment process for this exploratory study; we did not anticipate our findings would be exhaustive; however, we believe that they add to the understanding of this emerging domain. On the basis of our findings, we believe that individuals with higher technical knowledge and motivation to protect their privacy were under-represented. Finally, the strength of qualitative research is its ability to describe and understand both obvious and latent phenomena contained within the “thick descriptions” provided by interview data. Although our interpretation of these exploratory data is nongeneralizable, the use of in-depth interview methodology provides researchers with an appreciation of the complexity and context of this relatively new research domain. It should also be acknowledged that with every new innovative technology, the patterns of risk and security concerns may differ from those of ostensibly similar legacy systems [24].

The use of smartphone-based patient experience surveys provides new and exciting opportunities for health care providers to assess and improve the quality of health services. We conducted 24 semistructured interviews with smartphone users to explore the types of perceived risks that may exist when using smart surveys in the context-sensitive health services sector. The results demonstrate that the classical dimensions of perceived risk raised minimal concerns for the use of smartphones to collect patient service experience feedback. However, PAS risk associated with trust in the doctor-patient relationship, the reliability of the communication channel, and the risk of potential loss of agency over shared information may inhibit adoption. Conversely, altruistic motivations to contribute to health service quality for others may facilitate patients’ adoption of smart surveys. We conclude that barriers and enablers of adoption of novel technologies may change from sector to sector and should be further explored.