Refining the design of a smartphone application for people with chronic low back pain using mixed quantitative and qualitative approaches

Abstract Introduction A mobile application has the potential to involve people with chronic NSLBP in their rehabilitation. To refine the design of a smartphone application for people with chronic NSLBP using mixed quantitative and qualitative approaches. Methods We used a user-centred design approach involving people with chronic NSLBP and healthcare professionals (HCPs). We used a three-step methodology: developing consensus on the features, content, and design of the app; developing a user interface; and usability testing of the app and assessing users’ experience. Transcripts of interviews of users were analyzed by qualitative content analysis. Results A total of 18 people (aged 45 [23–53] years old) with chronic NSLBP, and 7 HCPs (aged 29.5 [25–55] years old) involved in NSLBP management were interviewed. The overall experience of using the smartphone eLombactif app was initially assessed. Then, with close-ended questions we evaluated users’ judgements on the content, its presentation and navigation. Finally, we asked for suggestions: “application content and functionality” and “content presentation” from participants regarding the use and development of the app analyzed by a qualitative methodology. Conclusions This study described how we refined the design of our application for people with chronic NSLBP using a qualitative and quantitative approaches. This methodology allows for deepening the knowledge of the needs and expectations of potential users by measuring their user experience. IMPLICATIONS FOR REHABILITATION Non-specific low back pain (NSLBP) is a major global public health issue leading to considerable economic cost and is primarily responsible for pain and disability. Mobile application has the potential to involve people with chronic NSLBP in their rehabilitation. This study described how we refined the design of our application for people with chronic NSLBP using a qualitative and quantitative approaches.


Introduction
Chronic non-specific low back pain (NSLBP) is defined as temporary back pain with no medical signs of a serious underlying condition (e.g., cancer, infection, fracture, spinal stenosis) [1] that has been evolving for more than 3 months [2]. For the general adult population, the mean overall prevalence of low back pain (LBP), which was defined as all prevalence regardless of prevalence period, was 31.0%. The mean point prevalence was 18.3%, and 1year prevalence was 38.0% [3]. This condition is one of the leading causes of disability and work stoppage among working-age adults [4]. Although 80-90% of patients recover within the first 2 months after diagnosis, the shift to chronicity is responsible for 60-80% of economic costs [5,6].
The results of chronic NSLBP management are better with different interventions combined, such as conventional treatments, physical exercise and cognitive-behavioural therapies (CBTs) [7]. Indeed, the American Pain Society and the American College of Physicians recommend coupling CBTs with active functional restoration exercises as part of interdisciplinary management [1]. These strategies appear to be effective for the factors affected by pain, such as health-related quality of life, depression and return to work [8].
Psychological factors such as fear of pain, fear of movement, catastrophism, and inadequate management strategies quicken the transition to chronicity [9][10][11]. Anxiety, depression, non-return to work and social isolation contribute to the progression of NSLBP symptoms [12]. Behavioural therapies aim to limit the negative impact of these factors on disease progression: physical reactivation, relaxation learning, stress management and coping strategies (such as positive self-affirmation, distraction) [11]. Some interventions may aim to extinguish painful behaviours by modifying pain perception via operative learning and re-learning techniques [13]. Extinction includes a gradual increase in the frequency of healthy behavioural phases over an extended period. This type of behaviour depends on the context (hospital, home, work, etc.) that patients find themselves in and that can be neglected in the absence of reinforcement [14].
Smartphone applications may be one of the most appropriate CBT tools to stimulate relearning and promote the sustainable development of healthy behaviours [15][16][17][18]. Indeed, they can create a favourable microenvironment for patients' care by giving them easy access to valid and individualized medical information, reassuring messages, self-management tools and monitoring of disease progression by the medical profession. Very few applications (apps) dedicated to chronic NSLBP have been developed [19].
According to the concept of user-centred design [20], end users must be integrated into the design of a computer tool from the beginning [17]. Therefore, to develop an app intended for people with chronic NSLBP, we conducted previously a qualitative study which gathered the expectations of potential users (people with NSLBP and health care professionals [HCPs]) as well as the obstacles and facilitators to use [16]. The main limitation of studying barriers and facilitators when using an app is the difficulty for users to project themselves into the future and imagine the potential use of the app. This prototype can give participants the opportunity to become more actively involved in the design and express their needs [16]. The main objective of this study was to refine the design of a smartphone application for people with chronic NSLBP using a mixed quantitative and qualitative approaches.

Development team and platform
A first version of the app (eLombactif) was developed by the team of the Physical Medicine and Rehabilitation Department of Clermont-Ferrand University Hospital, including 1 physician, 2 physiotherapists, 1 adapted physical activities teacher and a PhD student. This app uses the Android platform (min API 14) due to the expertise of our developer.

Development of the user interface
To maximize the effectiveness of an app, its design must include ongoing evaluation of user experience [20]. Even though there is a model that tries to predict the quality of the user experience, it is preferable to validate the user interface of an app empirically [21].
The target age group for chronic NSLBP users was 18-55 years, with an average age between 40 and 45 years. According to our previous study, the mean age of the target population was 41.9 ± 6.8 years [16]. In addition, participants expressed the need for computer support to be easy to use. Therefore, the main functions of the app should be defined and the user interface centred on these functions [20]. In this way, organizational logic can be more understood to reduce the time required to start using the app. Based upon these observations, the eLombactif prototype would include the main functions: (1) provide explanations on the disease and its treatment and (2) provide computer support for physical reconditioning sessions.
Determining in advance what type of user interface will produce a better user experience is difficult. Thus, the questionnaire for users consisted of three parts: (1) AttrakDiff questionnaire items on general usage experience [22], (2) close-ended questions on the user interface, and (3) open-ended questions on user expectations.
This type of questionnaire (general close-ended questions to open-ended questions) leads the respondent progressively towards a deeper reflection on the subject. Therefore, the interface will be altered as new versions of the app are released, evaluating it using a differential quantitative method. The interface is based on the AttrakDiff questionnaire, which allows for users to estimate their general impression of an app [21]. Each item was rated on a scale of 7 values between two antonyms (e.g., simple -complicated, uncontrollable -controllable). However, it did not allow for highlighting the reasons why users think that the app is uncontrollable or discouraging. Therefore, more in-depth questions on content, presentation, exchanges within the app and ease of use were added (Supplementary Appendix). These questions were based on the literature and the professional experience of the investigators who had published several qualitative and quantitative studies concerning physical activity promotion and NSLBP [23]. At the end of the questionnaire, open-ended Content is relevant to the chronic low back pain self-management program þ þ 2 Information about the disease and treatments is clear The vocabulary is adapted þ þ 4 Information sources are clear and well identified þ À 5 Scientific information is distinct from non-scientific information À À 6 Does the information in the application cover all your need? þ À 7 Do the skills mentioned in the application correspond to your needs? þ NA 8 Do the skills targeted in the app match my patients' needs? NA þ 9 The text is easy to read þ þ 10 Illustrations or animations are useful and relevant þ À 11 All graphics components are working correctly þ þ 12 Are you happy with the opportunity to give your opinion? þ NA 13 Is the way you can give your opinion to the developer about the app convenient for you? NA þ 14 Do you feel supported? þ NA 15 The app allows you to support your patients and monitor their progress NA -16 Is it easy to navigate in the application? þ þ 17 Easy understanding of orientation buttons þ þ 18 Easy navigation (round trip, return home page, application map) þ þ 19 The instructions for use are easy to understand þ þ 20 I did not encounter any difficulties when starting the application þ þ The item is annotated with a "plus" sign and considered acceptable if the median value is "Strongly agree" or "Agree", otherwise, with a "minus" sign for "neither agree nor disagree", "disagree" or "strongly disagree". "NA" means Not Asked.
questions were used to collect users' expectations [24]. Openended questions are part of a qualitative approach that aims to explore new aspects that are difficult to predict when designing a questionnaire [25].

User experience evaluation Study scheme.
We constructed a descriptive study protocol for evaluating the needs of people with NSLBP and for HCPs, with a pseudo-random survey following the World Health Organization methodological guide and the COnsolidated criteria for REporting Qualitative research (COREQ) guidelines [26]. The evaluation study of the eLombactif app by potential users was conducted with two groups of participants: people with chronic NSLBP and HCPs. Patients were those enrolled in the Lombactif self-management program, which is designed for people with chronic NSLBP not overcome with outpatient treatment. This 3-week pluridisciplinary rehabilitation program combines educational and physical reconditioning sessions at Clermont-Ferrand University Hospital. HCPs consisted of adapted physical activity therapists and teachers, physicians, and physical therapists involved in the management of chronic NSLBP in two university hospitals: Clermont-Ferrand University Hospital and Montpellier University Hospital.

Data collection and analysis.
After patients signed the consent form, the study and the eLombactif app were presented orally to people with NSLBP in the physical medicine and rehabilitation department of Clermont-Ferrand University Hospital and Montpellier University Hospital during a self-management session on physical activity adapted to the beginning of their stay (week 1). Participants then had weeks 2 and 3 to use this tool during physical reconditioning sessions or at home to identify any difficulties in using it. At the end of the third week, participants were asked to complete the patient questionnaire.
Data for Parts I and II of the questionnaire were processed by a co-investigator (A. A.). Median values were calculated for each item and represented by a Kiviat diagram. The Part II item was considered acceptable if the median was "Strongly Agree" or "Agree" and annotated in the presentation of results with a "plus" sign ("Strongly Agree" or "Agree"), otherwise with a "minus" sign (see Table 1). Data for Part III of the questionnaire were analyzed independently by two co-investigators (A. A. and I. F.) for the two centres. All responses were listed and compared, then grouped into categories and subcategories if similar. The responses were then independently reviewed and counted to rank them according to their recurrence. In case of disagreement, a third investigator was consulted (E. C.).
The HCPs of the physical medicine and rehabilitation departments were informed of the study individually. They evaluated the app before completing the HCP questionnaire which is nearly identical to the patient's one.

Recruitment.
Data saturation has attained widespread acceptance as a methodological principle in qualitative research. In the case of this preliminary study and the absence of baseline data on the number of participants to be included, we considered the flow of patients with chronic NSLBP through our department. The active annual patient caseload for the Lombactif self-management program is approximately 50 patients. Because the survey period was limited to 5 months, the number of patients that could be included was estimated at 23. According to the investigator team's experience, 3 of 4 people meet the main inclusion criterion (having a mobile device running the Android operating system, version � 4.1). How the study was conducted allowed for predicting a very high participation rate. Finally, the number of patients to be included was set at 15.
Patients of both sexes were included if they were age 18-55 years old, had chronic NSLPB and were registered in the Lombactif self-management program, had a mobile device with the Android operating system version � 4.1 and had the necessary computer skills. They were excluded if they had problems understanding or expressing written French.
HCPs of both sexes were included if they were a physician/resident, physiotherapist, adapted physical activities teacher, occupational therapist, nurse or psychologist; were experienced in the management of chronic NSLBP; had a mobile device under the Android operating system version � 4.1; and had the necessary computer skills.
Ethical data. University IRB approval and participant consent were obtained prior to recruitment and data collection. All participant procedures were performed in accordance with the ethical standards of the Local Ethics Committee (CPP Sud-Est VI: 2016/CE76) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Description of the "eLombactif" app
The main function of the app is to support physical exercise training. The first version of the app was organized as follows. First, the user is directed to learn about the pathology, risk and protective factors and treatments. The target audience was people with chronic NSLBP who completed the Lombactif self-management program and already had knowledge of physical exercise. After having understood the usefulness of physical activity, the app proposes that the user begin the practice of predefined exercise sessions. If necessary, the app contains the description of the exercises with a picture or a video recording. The user can also build their own sessions on the basis of the exercises.
More details and screenshots are available: Supplementary Appendix.

Population
The HCP sample consisted of 7 individuals (4 men), with median age 29.5 years (range 25-55). The patient sample consisted of 18 individuals (9 men), with median age 45 years (range 23-53), with various socio-professional categories. Among the 18 questionnaires distributed to people with chronic NSLBP, 2 were returned with incorrect data and 1 was not returned. Seven questionnaires were administered to HCPs and all were usable.

User experience
General impression of the experience of using the application The first part of the questionnaire gathered general impressions of the experience of using the app (Figure 1). People with chronic NSLBP and HCPs agreed that the aspects "Useful, motivating, controllable, clear and simple" were well developed. However, they could not comment on the elements of "socialization" with the app.

Users' judgement of the app content
The second part of the questionnaire clarified the users' judgement of the app content, its presentation, and navigation in the app. Most items received positive feedback from study participants ( Table 1). The median for item 5 for patients and HCPs was "neither disagree nor agree" and was similar for items 4, 6, 10 and 15 for HCPs. Blank box mean we don't assess the question.

Participants' suggestions
All participants in the patient group confirmed that they wanted to use this application and would recommend it to people they knew with NSLBP. Among HCPs, 5 of 7 planned to recommend this tool for physical reconditioning and sometimes with the follow-up of a physiotherapist. Table 2 presents the ranking of participants' suggestions with respect to the app's use according to the frequency of suggestions. All responses were grouped into two categories: "application content and functionality" and "content presentation."

Discussion
The methodology chosen for this research meets its objective: Refining the design of a smartphone app and the evaluation of the experience of its use by a mixed approach. These data confirm other recent scientific work published on the same topic [16,18,27,28]. The user-centred design model of app development includes close collaboration in the design and evaluation of this tool. This model recommends the focus on user experience, which in turn depends on the user interface. Indeed, the organizational logic of the simple app and the clear and intuitive design facilitates patient adherence to long-term treatment, here treatment for NSLBP. Otherwise, the user may be discouraged when the use of the app generates permanent discomfort, even if the content is good.
The first version of the app was well accepted by patients and HCPs. Among the first 100 users, 78 kept the app on their mobile devices. The comments of the HCPs mainly concerned the occupational therapy component and the motivational aspect; patients' expectations were directed towards the practical side of setting up the exercise sessions (timing).
Results of Parts I and II of the questionnaire showed that the app users could feel excluded and more socially isolated. However, this aspect was not raised in Part III and so will require further research on the nature, topics and importance of potential social exchanges between the patient and therapist and between the patients themselves.
The number of HCPs involved in the management of NSLBP in the two centres is about 25. According to the preliminary survey, the number of HCPs with a mobile device under the Android operating system is much lower as compared with patients. The sample size for HCPs was set at 10. A smaller number of participants from the HCP than initially planned prompts reflection on the need to develop a version of the app under the iOS operating system. This development will reach a greater number of users among HCPs and patients.
Finally, patient and HCP suggestions were quite different: for patients, they centred more on the exercises, whereas HCP suggestions were related more to add educational and motivational content.

Limitations
The app was developed on an Android platform only, which in itself was limited in offering the app to all patients. In our experience, 75% of patients use a smartphone running on an Android system. Nevertheless, we plan to develop the application on an iOS so that it can benefit all patients and HCPs.
Although this was a pilot study, clinical factors such as pain intensity, associated symptoms, anthropomorphic measures, and functional disability assessment could have been introduced to better understand the study population.
The adherence rates to apps for physical activity promotion significantly decrease within 3 months [29]. At this time, the app does not allow the HCP to support patients, monitor their progress and maintain the adherence so we hope to add this feature in the future.

Perspectives
The main advantage of quantitative evaluation of the experience of using the eLombactif app is to be able to make a discriminative judgement on its two different versions. This will allow us to focus on the less developed aspects. The prospects for future work are to update the eLombactif app according to participants' expectations and to set up a tool to evaluate the experience of continuous remote use.
Depending on the behaviour change taxonomy and the results of previous studies, the ability to monitor users remotely must be developed [16,27]. This development will meet the demand of patients and HCPs to act on the motivational aspect of disease management via a Smartphone app. Indeed, regular exchanges with health professionals will contribute to the reactivation of relearning and the extinction of painful behaviour [14]. However, monitoring can provide data on the evolution of patients' health status to design a study to evaluate the effectiveness of such an intervention in the long term (3 or 6 months).

Conclusion
Here we refined the design of a self-management tool for patients with chronic NSLBP (eLombactif app). The study allowed for deepening the knowledge of the needs and expectations of potential users and measuring user experience. The results of this study can serve as a reference for future adjustments and improvements to our app.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by CHU Clermont-Ferrand, Clermont Auvergne University. AA received grants for the MSc degree from Thuasne.