1
SMART SENSORS IN THE WORKPLACE
Views on Introducing Sensor-Based eHealth Tools to Improve Healthy Working
MASTER THESIS
MSc Supply Chain Management University of Groningen Faculty of Economics and business
RALPH ALEXANDER LINGAARD
2
ABSTRACT
Background: A leading cause for mental illness, musculoskeletal disorders and cardiovascular disease among desk-based employees are stress and sedentary behaviour at work. Sensor-based eHealth tools (SBET) could be utilised for self-assessment and goal setting to decrease absenteeism and increase employability.
Objective: This study examines the views of desk-based employees and healthcare professionals on introducing SBET and identify facilitators and barriers to better meet the needs of the users and providers of care when using SBET.
Methods: A multiple case study is used to collect data. Semi-structured interviews are conducted with desk-based employees and healthcare professionals (HCP). A hypothetical concept “HealthySensor” was presented in conjunction with the interviews.
Results: This study included 14 participants (average age 42 – 71% male) consisting of desk-based employees (n=10), occupational physicians (n=2), and experts in the field of labour participation and ergonomics (n=2). Overall, participants were positive about SBET, considering it useful to drive healthy working. Nevertheless, privacy, design and technical issues were identified as barriers for adoption. Conversely, ease of use and perceived usefulness facilitated adoption. Furthermore, next to advice from HealthySensor, participants believe that counselling and guidance from a professional will aid user engagement. In addition, the resources made available by the employer and management support are also identified for long term engagement.
Conclusions: Participants are positive about SBET and believe it to be beneficial for their health. Nevertheless, they argue that using SBET should be supported by HCP and their employer, advice should be personalised and guarding privacy is paramount. Future research should consider piloting HealthySensor to check whether results remain the same after hands-on use.
Keywords: Sensor-based eHealth tools, Technology acceptance, sustainable employability, barriers and facilitators
Supervisor: E.I. Metting Co-assessor: J. Riezebos
3
TABLE OF CONTENTS
ABSTRACT ... 2 PREFACE ... 5 1. INTRODUCTION ... 6 2. THEORETICAL FOUNDATION ... 8 2.1. WHAT IS EHEALTH? ... 8 2.2. OCCUPATIONAL ABSENTEEISM ... 82.3. OCCUPATIONAL SENSOR-BASED EHEALTH TOOLS AND DECISION SUPPORT SYSTEMS ... 9
2.4. ORGANISATIONAL CHANGE AND TECHNOLOGY ACCEPTANCE ... 10
2.5. VIEWS ON EHEALTH ... 11 2.6. SYNTHESIS ... 12 3. METHODOLOGY ... 13 3.1. RESEARCH DESIGN ... 13 3.2. CASE SELECTION ... 13 3.3. DATA COLLECTION ... 14 3.4. INTERVIEW STRUCTURE ... 15 3.5. DATA ANALYSIS ... 16 3.6. ETHICS ... 16 4. RESULTS ... 17
COVID-19 IMPLICATIONS ON STUDY ... 17
4.1. PARTICIPANT CHARACTERISTICS ... 17
4.2. WORK ENVIRONMENT OF THE EMPLOYEES ... 17
4.2.1. Type of workspace ... 17
4.2.2. Environmental hindrances ... 19
4.2.3. Office ergonomics ... 20
4.3. PERSONAL HEALTH ... 20
4.4. PREFERENCES &NEEDS FOR HEALTHYSENSOR ... 21
4.4.1. Attitude towards HealthySensor ... 21
4.4.2. Advice ... 21
4.4.3. Communication ... 22
4.4.4. Privacy and Safety ... 23
4.5. HEALTHYSENSOR ADOPTION ... 24
4.5.1. Individual characteristics ... 24
4.5.2. eHealth tool characteristics ... 25
4.5.3. Implementation climate ... 28
4.5.4. Innovation characteristics ... 30
5. DISCUSSION ... 31
5.1. COMPARISON WITH CURRENT LITERATURE ... 32
5.1.1. Attitude of employees towards SBET ... 32
5.1.2. Barriers and facilitators to employee engagement ... 32
5.1.3. Barriers and facilitators to HCP adoption ... 33
5.1.4. Peer contact and interaction ... 33
5.1.5. Design and technical concerns ... 33
5.2. IMPLICATIONS FOR THEORY AND PRACTICE ... 34
5.3. LIMITATIONS AND FUTURE RESEARCH ... 34
6. CONCLUSION ... 35
REFERENCES ... 36
APPENDICES ... 42
4
APPENDIX B–INTERVIEW PROTOCOL HEALTHCARE PROFESSIONALS ... 48
APPENDIX C–HEALTHYSENSOR SLIDESHOW6 ... 52
APPENDIX D–MEDICAL ETHICAL APPROVAL ... 59
APPENDIX E–INFORMATION SHEET &INFORMED CONSENT FORM ... 60
5
PREFACE
This thesis is written as a completion of my master’s degree in Supply Chain Management at the University of Groningen. First and foremost, I would like to give my deepest gratitude to my supervisor, Dr. Esther Metting. She was a guiding beacon of support and provided me with useful advice and introduced me to her network and stakeholders of the project. Next to that, I would like to acknowledge prof. dr. Jan Riezebos, as co-assessor of this thesis for his time and feedback. My thanks goes out to IMDI SPRINT for providing me access to their research project and willingness to answer questions. Next I wish to thank Paul Mulder, for trusting me with the HealthySensor concept and his explanations of it. Furthermore, I would like to express my appreciation to the participants of this research for taking the time to share their views and experiences with me. Many thanks to Marijke Altena for her feedback, sharing ideas, proofreading and moral support throughout the duration of this project. Next, I would like to thank Matthijs ten Doornkaat and Koen Kuipers for their valuable contributions to this thesis. In addition, I want to voice special appreciation to my girlfriend Eveline de Vries for standing by my side and her continuous support. Finally, I wish to express my gratitude towards my family and friends for their support and belief in me.
6
1. INTRODUCTION
An increasing absenteeism rate in the Dutch working population is a problem in recent years. Currently the average annual national absenteeism rate is at 4.4 percent, compared to 3.8 percent in 20141. There are three major causes for absenteeism, mental illness, musculoskeletal system complaints and cardiovascular disease. Together these three causes account for over 60 percent of absenteeism in the Netherlands2. In 2017, ten million days of absence were recorded due to stress at work, which already relates to 2.8 billion euro in absenteeism costs3. Currently, 45 percent of employees feel that measures should be taken against stress to reduce the possibility of mental illness induced absenteeism3. Hendriksen (2013)4 argues that employers benefit from keeping their employees healthy; a proactive policy on sustainable employability could reduce absenteeism costs and improve employee wellbeing.
A study into occupational sitting and health risks suggests sedentary behaviour is negatively associated with health and is a risk factor for developing musculoskeletal problems; however, the evidence is limited4–6. On average the Dutch working population spends seven hours per working day in a sitting position7. Occupations in which employees get little physical activity are legislators, scientific professions, transportation and trade occupations7,8. Therefore, desk-based employees pose a group for initiatives that reduce sedentary behaviour and aim for sustainable employability.
eHealth tools can provide an individual access to information about their health status by measuring and reporting on biometric signs. This allows individuals to devise interventions and improve their health9. A key component for success in sustainable employability initiatives is for employees to change their behaviour, use new policies and tools for healthy working. To achieve this, unobtrusive eHealth tools can be utilised to accomplish accurate and real-time monitoring of biometric signs in workspaces. E.g. wearable sensor-based devices are used as they ensure little user awareness10.
7
Ross, Stevenson, Lau, & Murray (2016)11 identified facilitators and barriers employees might have for adopting eHealth tools in government and healthcare systems. Nevertheless, academic research into the opinions, facilitators and barriers on introducing (SBET) into the occupational sphere is scarce. To gain insight into the adoption of SBET by desk-based employees and the views of HCPs on utilising biometric data of employees, both perspectives will be juxtaposed. Therefore, the following research question is composed:“What are the views of desk-based employees and healthcare professionals on introducing sensor-based eHealth tools in the workplace?”.
A multi case study is performed to gain an initial understanding of the feasibility of introducing HealthySensor in the occupational sphere. The data is collected by conducting in total 14 semi-structured interviews with both employees as well as HCPs. This research contributes to literature by elucidating the views of employees and HCPs on introducing SBET in the workplace. Therefore, increasing the understanding of the factors influencing adoption of SBET. The research offers a practical contribution by providing IMDI SPRINT with knowledge on how to develop their concept to drive healthy working. This allows managers to devise strategies for introducing SBET into the workplace.
8
2. THEORETICAL FOUNDATION
In this section the theoretical foundation is discussed. It commences with an introduction of eHealth, after which occupational absenteeism is explained. Next, SBET and decision support systems (DSS) are discussed. The next section addresses technology acceptance and organisational change. Thereafter, the views of employees and physicians on using eHealth tools is discussed. The theoretical foundation concludes with a synthesis of the discussed literature and a conceptual framework.
2.1. What is eHealth?
eHealth is a quickly developing and highly debated research topic in recent years. Different authors propose several definitions of eHealth. The World Health Organisation states that utilising ICT in support of health and health-related activities is the definition of eHealth12. This broad definition fails to capture that eHealth is also used to replace or supplement health-related activities. Furthermore, eHealth can be split up into two categories: TeleHealth and mHealth13. Audio/video communication between a physician and a patient, who for example lives in a remote area, is considered Telehealth. Whereas mHealth is the use of mobile technologies, allowing for real-time access to information and services13,14. eHealth is seen as an umbrella term for the aforementioned categories. Therefore, eHealth can be explained as the conversion of healthcare systems and patient-related processes shifting from manually performed tasks to the tasks enhanced by or replaced with Web-enabled systems and processes15,16.
2.2. Occupational absenteeism
In the period between 2013-2017 the absenteeism rate increased for both men and women in the Dutch working population due to mental illness, from 26.4 to 32.7 percent, as a percentage of total absenteeism. The proportion of absenteeism due to complaints from the musculoskeletal system has decreased from 33.3 to 28.6 percent in the same period. This is due to the decrease in the total number of absenteeism days for all disorders2. Nevertheless, both mental illness and complaints from the musculoskeletal system account for over 60 percent of total absenteeism of the Dutch working population. The third cause for absenteeism is cardiovascular disorders.
9
with insufficient resources, little recovery or mismatches with personal characteristics lead to a stressful work environment21. Additionally, Koldijk et al. (2016)21 argue that prolonged exposure to stress can lead to long-term health consequences.The second and third major causes for absenteeism are complaints from the musculoskeletal system and cardiovascular disease. Work-related musculoskeletal disorders are caused by mechanical overload, repetition frequency, exposure time, posture and accidents in the workplace6. Imbalance of nutrition, physical inactivity, disturbed sleep cycles or stressful environments at work, contribute to developing cardiovascular disease 22. Van Uffelen et al. (2010)5 identified that occupational sedentary behaviour (prolonged time sitting in one place) has negative associations with health. A study comparing sedentary behaviours of more than 200,000 adults (45 years and older) found that there was a 15 percent increased risk of all-cause mortality with people sitting less than four hours in a day compared to people who sit between 8-11 hours a day. Furthermore, an increase of 40 percent was found with people sitting over eleven hours a day23,24. In contrast higher levels of occupational, recreational and total physical activity are associated with a reduced risk of all-cause mortality8. Furthermore, a study into the sedentary behaviours of Dutch workers found that employees sitting for prolonged periods at work did not compensate such behaviour by sitting less outside of work. To reduce health problems, interventions to reduce sedentary behaviour should take place at work, when traveling to and from work and in time outside of work7.
Mental illness, cardiovascular and musculoskeletal disorders caused by occupational hazards are grave issues that result in large absenteeism numbers. Stress, physical activity, sedentary behaviour, posture and the work environment are factors that can be quantified and measured by sensor-based technology, which might reduce the level of absenteeism21.
2.3. Occupational sensor-based eHealth tools and decision support systems
eHealth is to the public most recognisable as the many health and fitness applications that are available on mobile consumer devices. Mobile phones and smart wearables use sensor-based technologies to measure personal activities and biometric signs, such as heart rate and motion among others10,25. Multiple scenarios can be thought of where humans being monitored with smart wearables integrated in an Internet of things (IoT) scenario can be utilised. An interesting way of using sensor-based devices is to improve the quality of life26. Developments in the IoT make it possible to monitor people using sensor-based wearable or non-wearable devices outside of a clinical setting10. Using SBET enables individuals to monitor their health in real-time27,28.
10
to try it25. Examples of eHealth tools utilised in the occupational sphere are wrist-worn devices and smart office chairs. Wrist-worn technologies, such as smart watches or activity trackers collect data on the number of steps taken, sleep patterns, heart rate and number of calories burned among others. Next to that, wearable eHealth tools are emerging as important devices for long term health monitoring28. An office chair fitted with sensors in its seat surface and backrest can be used to measure sitting behaviour, musculoskeletal discomfort and correct sitting behaviour by providing tactile feedback by sending pulses to the user30. A smart chair may be used to reliably monitor individuals performing regular desk-based tasks in a non-intrusive manner and could therefore be used to monitor an individual’s health for a prolonged period of time29. Nevertheless, the feedback signal led to insignificant changes in sitting behaviour30.The health and activity data collected from users by the eHealth tools is often stored in a cloud-based environment. That data is processed, and analysis is performed to understand patterns in health information31. The Big Data yielded from the tools is best processed and analysed with a DSS. The processed data could provide a holistic view towards health hazards of people from their basic physiological data31. The cloud-based infrastructure can be used to distribute the results of the analysis to the users or other stakeholders10,25. Thus, when an individual has access to collected data about their health status, they can monitor their health condition and lifestyle. Thus, allowing individuals to self-assess and perform interventions27.
Koldijk et al. (2016)21 argue that workers are relatively flexible in executing their tasks, therefore they could improve their own wellbeing at work. They researched how pervasive and unobtrusive sensors could measure workers and provide insight into their health by utilising a system of support applications, so they can manage their health and act immediately. The results of their study showed that users are only interested in collecting data that is necessary for achieving their specific goals and that specific behavioural change techniques should be personalised to motivate or trigger the employee to take action21.
2.4. Organisational change and technology acceptance
11
The Technology Acceptance Model (TAM) is an information systems theory that describes the acceptance and usage of technology by individuals. The TAM model is presented in Figure 2.1. TAM identifies the Perceived ease of use (PEOU) and perceived usefulness (PU) as antecedents to determining attitude and intention to use a technology34. The model argues that people use a technology depending on their PU, which is defined as “the user’s perception of the degree to which the system will improve his or her performance in the workplace” (Venkatesh & Davis, 1996:452)35. The PEOU is defined as “the user’s perception of the amount of effort needed to use the system” (Venkatesh & Davis, 1996:452)35. Users base their PEOU on self-efficacy, before they actually used the technology35. As such, direct hands on contact with the technology is not necessary for users to determine their intention to use the technology or not. Enhanced self-efficacy beliefs could stimulate acceptance due to users feeling more comfortable about the technology. Next to that TAM is used in multiple sectors as foundation for introducing technology36. The acceptance of technology should be conducted along with qualitative research to delve deeper into underlying reasons for system use37. This will allow managers to devise a strategy to implement the technology in a specific context.FIGURE 2.1
Technology Acceptance Model
2.5. Views on eHealth
12
autonomy in health management, bottom-up development of an eHealth system generates enthusiasm and commitment. Next to that, design, technical concerns, privacy, PEOU and whether it is imposed on employees were identified as being critical for eHealth acceptance38,39.The views of physicians on adopting eHealth technology in their practice is studied by de Grood, Raissi, Kwon, & Santana (2016)40. They identified that the most regularly mentioned barriers and facilitators of adopting eHealth technology by physicians are a threat to clinical autonomy, liability issues, and training & support. Some physicians believe that their ability to make autonomous clinical decisions will be interfered with due to the information provided by the eHealth tools38,40. This is endorsed by Walter & Lopez (2008)41 who found that threatened professional autonomy is negatively related to the PU and the intention to use eHealth tools. The barriers erected were related to expenses of maintaining eHealth devices and financial incentives would facilitate adoption40. The third major issue concerns training and support. The degree of IT support and training and education in using the tools has a substantial influence on accepting eHealth technology38. Training should be tailored to a physician’s technical skills or be given to an assistant as it would be extra work for them to learn how to work with the eHealth tool. Next to that, the lack of ICT skills and a large workload is determined a barrier to eHealth utilisation40,42. Conversely, research proving the utility of the eHealth tool and the tool being able to generate data that assists physicians in detecting adverse events are facilitators40,43.
2.6. Synthesis
13
FIGURE 2.2Conceptual Framework
3. METHODOLOGY
3.1. Research design
The purpose of this exploratory study is to identify opinions, expectations, barriers and facilitators of introducing the HealthySensor concept in firms, from the point of view of both desk-based employees as well as HCPs. The aim is twofold. First, to identify why employees’ behaviour does not change when they receive feedback on it and to determine if and how HealthySensor needs to be further developed to increase adoption by desk-based employees. Secondly, this study aims to explore how HCPs might utilise a system like HealthySensor and to determine their expectations and requirements for such a system. The employer’s point-of-view is not covered in the research, due to time constraints. The research question can be answered by performing qualitative research. A case-study design is the most suited for qualitative research for the following reasons: First, case studies are typically used for examining unexplored phenomena. Second, it allows for studying phenomena in its natural setting, through observing actual practice. Third, it presents the opportunity to answer why, what and how questions with a relatively well understanding of the studied phenomenon45,46. As this study aims to gain insight from multiple perspectives, a multiple case study design is performed47. The study utilises a few in-depth interviews to obtain a substantial amount of detailed data48. The study is performed between February 2020 and July 2020. The data collection is executed in March and April 2020.
3.2. Case selection
14
perspectives introducing SBET in firms. Next to that a multiple case study decreases researcher bias and enhances external validity45,49.In a case study selecting the right cases is paramount. Thus, boundaries that define what can and cannot be studied are determined45. The cases are selected based on theoretical replication logic. The views, barriers and facilitators are expected to differ between cases as each actor group has different needs and requirements for such a system. Therefore, contrasting results between cases are expected which is an indicator for theoretical replication46. The employees are selected through three sampling methods. Firstly, the employees are purposefully selected based on their occupation. The inclusion criteria are that the employee should work at an organisation that classifies as one where sedentary behaviour is prevalent, such as legislators, transportation and trade professions. Furthermore, the occupation should be known for its prolonged periods of sedentary behaviour. Secondly, acquisition through the network of the researcher. Finally, a snowballing sampling approach is used. Interviewees identified individuals as possible participants for the study, which was useful for accessing difficult to reach respondents50. The HCPs and experts were selected utilising the same three sampling methods.
3.3. Data collection
The data for this study is collected through the means of semi-structured interviews. It is used to allow for follow-up questions and to delve deeper into the opinion of the participant on a specific subject46. The role of the participants is to provide the researcher with their unique experiences and opinions. Therefore, during the interviews it was imperative to not impose understanding of prior literature or opinions of the interviewer on the participants51.
The protocol for employees was standardised, thus ensuring consistency of topics allowing for coding and comparison per case. The protocol for HCPs had a general structure; however, was deviated from to specifically ask questions relating to HCP’s expertise. The protocol evolved during the process to reduce ambiguities. In addition to the questions in the protocol, follow-up and probing questions have been asked. Summarising answers were used to assess whether the answers were understood correctly, therefore increasing trustworthiness of the data. The protocol was not shared with the participants before the interview as the research required their initial response. The protocol was pilot tested beforehand by interviewing an acquaintance and was adjusted accordingly.
15
3.4. Interview structureParticipants in this study had no access to HealthySensor, as it is still being developed. Their opinions were based on a hypothetical concept of HealthySensor. Appendix A presents the interview protocol used for the employees and Appendix B presents the generalised interview protocol for the HCPs. The interviews were conducted in combination with an online slideshow presented in Qualtrics in which the sensors are elaborated on with a description of their use and an image to provide a visualisation. The slides are presented in Appendix C. First the interview structure of the employees is presented in Figure 3.1, after which the interview structure of the HCPs is explained.
FIGURE 3.1
16
In step 3 of the interview, the employees were shown a hypothetical case in which they were presented with a surrogate point of view. When an employee could not think of an answer concerning their work environment, the surrogate case was applied to gauge their view on the subject from another perspective.The interviews with the HCPs and experts started with introductory questions relating to their specific field, after which they were asked about their previous experience with sensor devices. The same Qualtrics slideshow was used to introduce the participants to the sensor-based devices and the HealthySensor concept. The surrogate point of view was discussed, nevertheless it was not used in the interviews with the experts and HCPs. After discussing the concept of HealthySensor, they were asked how they would utilise such a system in practice, what kind of information would be useful for executing their tasks and if they would consider advising employers to use such a system and why.
3.5. Data analysis
The transcripts are coded and analysed using ATLAS.ti. The data is coded according to the coding scheme suggested by Gioia et al. (2013)51. Thus, the transcripts are coded individually to identify the most important concepts. The coding was an iterative process. First, open coding is used to extract relevant quotes from the transcripts. Secondly, axial coding is used to deductively group and link quotes. Thirdly, the axial codes are aggregated into themes to analyse patterns. The findings were analysed per category and similarities and differences were identified. The code tree is presented in Figure 4.1.
3.6. Ethics
17
4. RESULTS
The results section firstly provides an overview of the research sample and secondly the qualitative results from the multi-case study. This section is structured according to the main research topics, that are presented in the coding tree (Figure 4.1). The code tree is devised of the qualitative results and reflects the opinions of the participants. Section 4.2 describes the current work situation of the employees. Next, the views of employees and HCPs on their own personal health are discussed. Section 4.4 discusses the preferences and needs for HealthySensor. Finally, section 4.5 presents the results on factors influencing adoption of HealthySensor by employees and HCPs.
COVID-19 implications on study
The data was collected between March-May 2020. In this period, the COVID-19 pandemic occurred. Containment measures significantly influenced the study. Due to the containment measures, the interviews were primarily conducted via digital communication channels. Second, the pandemic significantly influenced the availability of HCPs. They are part of the population that is working harder due to the ongoing circumstances. Thus, many potential participants indicated that they were too busy to participate in the study. The results relate to the work environment pre-coronavirus.
4.1. Participant characteristics
This research included 14 participants (average age 42 – 71.4% male, 28.6% female); Table 4.1 provides an overview of the participant characteristics. The sample consists of desk-based employees (n=10), practising occupational physicians (n=2) and experts in the field of labour participation and sedentary behaviour (n=2). IMDI-SPRINT identified occupational physicians as important actors for introducing HealthySensor, therefore emphasis was placed on the acquisition of occupational physicians for the HCPs’ case. Next to that medical directors and experts in the field of sedentary behaviour and labour participation were reached out to. All but one participant was approached by e-mail; one participant was approached through telephone communication.
4.2. Work environment of the employees 4.2.1. Type of workspace
18
FIGURE 4.119
TABLE 4.1Participant Characteristics
Position Type of organisation Sex Type of interview Reference
Traffic controller Public transportation M Video call E1
Traffic controller Public transportation M Face-to-face E2
Project manager Medical diagnostics M Video call E3
Manager application management Medical diagnostics M Video call E4
Information manager Medical diagnostics M Video call E5
Procurement officer Government agency M Video call E6
Epidemiologist Hospital F Video call E7
Epidemiologist Hospital F Video call E8
IT-manager Medical diagnostics M Video call E9
Communications officer Public transportation F Video call E10
Occupational physician i.a. Health and Safety service M Video call OP1 Occupational physician i.a. Health and Safety service F Video call OP2 Professor of pain, revalidation and
labour participation
i.a. University M Video call EX1
Ergonomist i.a. Health and Safety service M Video call EX2
4.2.2. Environmental hindrances
20
4.2.3. Office ergonomicsThe traffic controllers have specialised 24-hour chairs that are equipped with specialised neck, back and armrests and can be adjusted in different ways. Next to that, they have sit-to-stand desks and multiple screens to perform their duties. Six other employees mention that they use a regular office chair. Two employees indicate that they have a specialised chair that has been adjusted by the occupational health and safety service. The findings indicate that most employees can adjust their desks in height, either electrical or manually. Nevertheless, including the traffic controllers, four employees have sit-to-stand desks.
4.3. Personal health
All employees believe that sitting for prolonged periods of time is bad for their health. Additionally, they all believe that their sedentary behaviour could be reduced at work. The traffic controllers argue that it is difficult to just walk away from their stations as they should always be available for calamities during their shift “We have an eight-hour shift. There is no official break in that shift” E2. By contrast, the managers argue that a major part of their working day is spend sitting in meetings or answering mails; however, they also have to walk a lot from conference rooms, back to the office or move to another location for a meeting. Nevertheless, most of their time is spent in a sedentary position. Also, more than half of the employees indicate that they easily take on a sitting position that is bad for their backs and experience backaches as a result.
Employees indicate that is it difficult to reduce sedentary behaviour. Nevertheless, some employees engage in standing during meetings or when reading e-mails, walking while having lunch or visiting a colleague instead of e-mailing, to reduce sedentary behaviour at work. In their leisure time employees try to compensate for a lack of physical activity at work by running, cycling, walking the dog, doing sports and working out among other things. Conversely, (2/10) employees indicate not to exercise at all and have no incentive to change their current behaviour. Furthermore, (7/10) employees mention that they are consciously engaged with eating healthy, by contrast (1/10) employee admits to rarely eating healthy due to irregular shifts. (2/10) employees do not mention their diet at all.
21
4.4. Preferences & Needs for HealthySensor4.4.1. Attitude towards HealthySensor
All employees mention that they are curious to test HealthySensor and are interested to see how it works and which results and advice it provides in practice. The results indicate that their attitude towards HealthySensor was formed by their previous experience with health monitoring, mental illness, musculoskeletal disorders, affinity with technology or their type of job. (4/10) employees mention a history of back, neck and wrist aches and believed that the system could assist in reducing those complaints. Specific points of scepticism were privacy concerns and the system’s ability to provide useful personalised advice. Several employees frequently use fitness and health applications on their device. They perceive these tools as helpful in monitoring their physical activity. Nevertheless, some employees have previously worked with software that locked their screens, forcing them to take a break or do some exercises. They perceived this as vexing, hence they stopped using it “My experience from that makes me a little sceptical, I tend to think it would irritate me quickly” E7.
Employees perceived HealthySensor as useful depending on the type of job it is utilised in; however, opinions were scattered. All employees believe that the system could best be utilised in an environment where people sit for most of the time or spend their time continuously typing. E8 and E5 indicate that it would be interesting for them “I think this is also something that suits me, because I only have sedentary work and I spend a lot of time at the computer typing or using the mouse” E8. Conversely, E1 and E2 argue that it would be less suitable for their type of occupation, since they have a more dynamic occupation. In addition, E3, E4 and E9 believe that their occupation is dynamic “Because of my flexibility, I just can't sit on it for a week. Maybe longer” E3. As such they believe that measurements do not provide a full image.
4.4.2. Advice
All participants believe that HealthySensor is useful for self-assessment. The employees argue that it provides insight into how long they are seated and in what position they are sitting. The results of the measurements will raise awareness into sedentary and breaking behaviour and also physical activity. The system allows setting goals and measuring progress. Next to that, advice could trigger users to critically review their breaking behaviour and potentially result in people moving more often.
“It's going to give insight; you see that it triggers in some way. Sometimes the body of thought is very different from reality” E9.
22
or recovering from burnout to take breaks more often and avoid overstrain. Also, E8 argues that she finds it difficult to listen to herself and an external trigger would motivate her to act. Both OPs and EXs share similar beliefs as the employees.The advices should be tailored to the individual and should not only consist of feedback in the form of facts or statistics. E7 argues that the advice should address a problem or stressor and propose several exercises the user could perform that are useful for addressing those issues “I don’t want a list of facts, that is of no use. It must be meaningful” E7. E9 mentions that he would like to see a benchmark comparing people with a similar age and occupation with the user. OP1 argues that tailor-made advice would be useful for secondary preventive care. He argues that someone reintegrating with back problems needs tailor-made advice as opposed to general advice.
Furthermore, people believe that the advice should be feasible. E5 argues that he had a similar experience years ago, where he had a periodic break of two minutes in which he needed to stand and perform simple exercises. He believed these short breaks to be surprisingly useful and feasible. By contrast, both OPs argue that it is difficult to change the environment as it is radical and expensive.
(8/10) employees argue that HealthySensor would be very useful for gathering evidence on a suboptimal workspace. The data from the environment sensor can substantiate a feeling that the environment causes headaches or drops in productivity “This will show you objectively whether you work in a healthy environment or not” E4. Next to that E5 argues that it could also be used by employers to start a discussion on occupational health and wellbeing when he sees that the group of employees has difficulty in exercising.
4.4.3. Communication
Participants indicate that the advice of HealthySensor should consist of three components. Firstly, alert notifications to indicate a change of sitting position or a need to take a break. Secondly, aggregated reports that provide insight into patterns over a longer period. Finally, the user should have real-time access into their health status. The results indicate that overall, advice from HealthySensor should be easy to read and presented in understandable language. E10 and E5 argue that long pieces of text will not be read, and that short messages and icons would be easier. E6 adds to that by mentioning that advice should be visualised and to-the-point to improve clarity. Using the wrong kind of formulation could backfire and cause the user to stop using the system.
23
noticing it anymore. The employees argued that reminders of notifications are necessary, as they don’t always have the opportunity to immediately act on advice. An option to temporarily mute alerts is suggested for situations in which a sound, buzz or tap is not desired. The health data should always be accessible and should provide information on the user’s current health status. Dashboards like Apple’s Health app or Microsoft’s MyAnalytics are examples. This allows for users to access it at their own discretion. Real-time information should also present a trendline to show the user how much deviation there is from the base-situation. OP1 argues that the system allows for real-time feedback, this is something an OP, employer or colleague is not able to do.“I would appreciate it if you can access it at any time, to actually have a kind of interactive app” E8 The aggregated reports should preferably be accessible via the dashboard in a mobile or web-based application. Participants are conflicted on the frequency of the aggregated reports. (5/10) employees mention preferring weekly reports, while (3/10) employees prefer a daily report. By contrast, E4 and E9 argue that a monthly report will provide more insight into trends.
4.4.4. Privacy and Safety
Employees were very concerned regarding the privacy related issues of HealthySensor. All participants believed that the personal data should be securely stored, such that people with ill intentions could not access it from inside or outside of the organisation. Furthermore, how data is stored should be made transparent to users. In addition, it is important to communicate what measures are in place to prevent data leakages or breaches. Several employees identify the sharing of data with the employer as a risk and are uncomfortable with it. Most of the employees do not want to share personal information if it is not anonymised or aggregated on group level; it is perceived as only for their eyes.
Nevertheless, some participants do not mind their employers having insight into their personal data. E1 argues that everything he does at work is already being monitored. By contrast, employees argue that the data from the environment sensor needs to be shared with the employer. They believe the employer should act based on those results. Furthermore, (4/10) employees feel that data could be shared incidentally with an HCP, after explicit permission of the user “If I actively go to a healthcare professional with a problem, they should be able to see everything” E8.
24
4.5. HealthySensor adoption4.5.1. Individual characteristics Short-term priorities
Employees believe that their short-term priorities such as meetings, deadlines, calls or concentration on a task are more important than to immediately act on advice provided by HealthySensor. They believe that often there is only one excuse; something needs to be finished. As such they believe that short or easy to perform activities are more likely to be executed. Most employees believe they do not have the time to walk for five minutes. E7 argues that even when she receives a feasible advice such as ‘stand up for a couple of minutes’, she will most likely not do it because typing while standing is uncomfortable. Some employees mention that they believe that their immediate tasks are more important; however, the system creates awareness to take action as soon as possible.
“When you are in a train of thought, you know you have to take a break physically, but mentally you really have to continue” E4
Knowledge and beliefs
25
4.5.2. eHealth tool characteristicsPerceived usefulness Smart chair
All employees are positive towards the chair and mention they would like to try it in practice. The chair is believed to be useful for indicating a bad sitting position or for providing alerts when someone is sitting for too long. Several employees believe their sitting position is bad and believe this chair can assist in changing sitting behaviour “When I get very tense, I crawl into my screen and no longer have a relaxed posture. I find it difficult to determine the best posture. So, I would really like a chair that can help with that” E8. A drawback for the chair is that is difficult to use by people with flexible workspaces. As such the chair needs to be adjusted every time someone sits on it. The managers indicate that they often move between rooms or locations. Therefore, every location should have this type of chair. If not, they worry that they do not sit long enough on the chair to measure their sedentary behaviour accurately, as mentioned by E3 in §4.4.1.
The OPs believe that people quickly adopt a bad sitting habit, the chair could increase awareness and prevent or allows people to change their sitting behaviour. EX1 argues that the operating system of the chair is based on the outdated model; sitting for a long time is bad and when you sit, you must sit well. He argues that the interpretation of data is difficult; however, the tools provide only feedback on the current situation.
Activity tracker
Most of the employees see the wearable as a useful device for measuring and providing insight into physical activity, heart rate and sleep. The employees that already use a wearable indicate that they like having real-time access to their health data. The OPs and EXs agree with the employees. EX2 also thinks coupling the wearable with a desk to automatically rise when prompted to would be a nice addition. E8 perceives it as a device that is not specifically used during working hours and is reluctant to use it due to privacy reasons. E2 argues that he does not like something on his wrist when he is sleeping.
26
substantiation of the issue. Thus, to focus on what will influence the issue and how to get or keep the employee healthy “This is of course definitely interesting with back problems or RSI complaints, but you can also imagine that it helps people with psychological complaints” OP2.Smart keyboard
The employees are divisive on the usefulness of the keyboard. Some perceive the keyboard as useful, while others think of it as unreliable. Half of the employees are curious about the keyboard and would like to use it. “I didn't know that you can do such a measurement on a keyboard. That is rather enlightening” E6. The keyboard provides insight into their typing behaviour and allows them to think about why their performance is increasing/decreasing during the day and devise interventions to regulate it. On the flipside employees are afraid that their behaviour or type of work conflicts with the accuracy of the measurements.
OP2 believes that if it functions as described it would be a very useful tool; however, argues that it should also be ergonomically and exclude the numerical part. OP1 perceives information from the keyboard as beneficial; however, adds that employees could have objections against using the keyboard. “From my position as a doctor, I don't see that many disadvantages to get feedback on computer use”
OP1. EX1 argues that gathering data from the number of keystrokes per minute would be very interesting for people with unidimensional work; however, the less unidimensional the job becomes, the more difficult it gets to use that data
Environmental tracker
The employees all believe this to be the most useful sensor to yield information on large groups and to identify stressors “I think temperature and lighting are the biggest disruptors at our office” E2. In addition, it allows for optimising the workspace to the user’s preferences “You will be able to make individualised adjustments. At the desired temperature, sound and light” E4. Conversely, E7 does not see the point in using it “I can't imagine that thing measuring things my body doesn't detect. I wonder if it adds much to what my biological sensors already tell me”
EX1 argues gaining insight into the environment allows for devising interventions. “We know that with a high CO2 content, concentration often decreases. A high temperature causes slack and too cold is also not good. Humidity also causes various complaints”. Nevertheless, OP2 argues against using the environmental sensor: “My experience is that if one is dissatisfied at work, people will complain about the indoor climate first. This is a way to identify their dissatisfaction”.
HealthySensor
27
potential psychological illnesses or musculoskeletal disorders. In addition, Both OPs and EX2 argue that they could utilise sensor data to provide guidance to counsel on physical activity and interpret eHealth data “It gives direct feedback to people and we use that feedback, because to be able to coach well you must of course know what is going on” EX2. They currently apply wearables for, measuring an employees’ goals and improving on them. To achieve effective counselling for achieving a healthy lifestyle, insight into the degree of physical activity, the working environment and sleep of an employee is deemed necessary.“If you cannot find a substrate, for pain or fatigue, it becomes very difficult to be able to give good advice. Then it would certainly help if you developed sensors that can measure that in some way” OP1 The chair, keyboard and environment sensors are deemed interesting for utilising as preventive instruments. OP1 mentions that he would use the wearable tailored to an individual. He argues that insight into physical activity and sleep is only necessary for employees in a vitality track or when he wants to motivate someone to get more physical activity or sleep better. This seems to suggest the data yielded by the wearable will be used to evaluate the progress in a vitality track.
“If you give someone sleep advice, it is nice that a wearable can measure whether it also has an effect or not” OP1
OPs argue that reporting of findings by the system should be short and simple; however, taking care not to lose important information. The data should be available before consult or be brought by the employee. Preferably, the report should be readable within two minutes. EX1 describes the environmental sensor as being the easiest sensor for interpreting results as they provide objective information on the environment, that could be cross-referenced with baselines of optimal working environments.
Design and technical concerns
28
system likely cannot take these aspects into account and even if it could, it would still not work due to the limited number of measure points. As such, he believes that the sensors will measure stressors accurately; however, it should also be able to translate the data into something usable. While it would not be possible for individuals, he argues that epidemiologically it should be possible. Drawing far reaching conclusions on an individuals’ health should be done meticulously and based on literature. According to EX2, there is no evidence in literature that directly substantiates sitting, environmental factors and typing behaviour to an occupational disease.“Based on how someone sits in their chair or the environmental factors at work or how often they type, you really cannot make a link to burnout yet another occupational disease” EX2
The wearable is perceived to yield untrustworthy data on sleep. OP2 argues that a slower heartrate is easily mistaken for sleep. She experienced deviations with several wearables in measuring sleep. The wearable identified a moment of rest on the sofa as a deep slumber. In addition, EX1 questions the accuracy of commercially available wearables, he perceives the data accuracy as mediocre at best. This is refuted by EX2 who argues that wearables have become more accurate in recent years and can provide accurate and trustworthy data. In his practice it is important to distinguish when someone is walking or sitting or to determine how much the heartrate increases when he is exercising. For those purposes absolute accuracy is not necessary “It doesn't come down to five heartbeats per minute more or less, so we think that bandwidth is acceptable” EX2.
The data generated by the smart keyboard is perceived to be of poor quality due to its inability to distinguish labour productivity from idle behaviour “What does that say about my labour productivity? I may be very productive; I just don't type as fast or rewrite more often” E2. Employees can also be productive while not operating a keyboard, when using a mouse, calling or being in a meeting. The keyboard is deemed very usable for detecting typing mistakes and decreases or increases in typing speed. Nevertheless, EX2 argues determining labour productivity based on decrease in typing speed or an increase in mistakes is drawing conclusions from assumptions and could yield questionable results. 4.5.3. Implementation climate
Available resources
29
means you need to have time to move” before employees can fully commit to HealthySensor. E8 believes that employers currently do not facilitate enough to improve workspaces and stimulate employees to exercise “I think much can be done about a workplace, such as standing desks or bicycles. There are a lot of possibilities that aren't really facilitated at all”.Leadership engagement
The findings indicate that the management of the firm should actively engage and voice support for employees using HealthySensor “The employer should facilitate this, because it is ultimately about the health of your employees” E10. Next to engagement, the employer could also have a monitoring function “As a manager you can also ask did you do that? Such that it is not pedantic but stimulating” E5. Furthermore, training and education should also be provided, otherwise people will use it wrongly.
“When you don't get it explained, you won't use it like intended. Then it’s just a chair, may be another black chair too” E9. Instead he argues: “When you start using it, there is someone from the company, or a champion of your department who explains your workplace”.
Peer contact / Interaction
Both OP1 and EX2 believe that only providing sensors do not provide enough incentive for people to change their behaviours. “Because they still get a bit of professional guidance. Only a smartwatch is not enough” EX2. Instead, introducing the technology well and supervision is perceived a key determinant for changing behaviour “You also have to help people to interpret those results and, to guide them to another behaviour. I think, the more personal guidance from a professional, the better”
EX2. OP2 adds that “someone supervising and providing feedback can compel users to act”. That view is also supported by the employees. All employees believe that such a system works better with groups of people and argue: “You encourage each other to keep moving. You can think of a group initiative” E6. Some participants argue for a competition element to stimulate exercise. This is already done at the department of E9: “We try to trigger each other. At the end of the day everyone has to come in and if you have less than five thousand steps, you’ll have to pay for dinner”. By contrast, E2 is concerned and states: “Absolutely not. Then you get differences between colleagues. We are a very small, but very close group, there should be no competition”.
30
provide evidence on an employee’s health and she will be able to cross-reference it with the story from the individual to determine whether he/she is telling the truth. “That is an employee’s story, if it is substantiated with sensors from the chair or from the environment. We can look in a targeted way, about what should be influenced now to get or keep this employee healthy” OP1Sustainable employability initiative
Several employees believe that HealthySensor should be part of an overall sustainable employability initiative. “You have to make an overall program for it as a company. This must be a theme where the firm will help employees to become more vital” E3. Firms should address more issues concerning vitality at work. “It means you have to do much more than just using sensors. That means you have restaurants where snacks are no longer available, and you have to introduce mandatory walking at noon” E4. Another employee believes that first a firm should take care to optimise their current work environment as much as possible. “If the working environment does not show at all that it promotes anything of health, then I wonder if you should immediately use sensors” E8.
Next to that, employees believe that firms should offer services to increase the likelihood of following advice “If they see okay my employees move less, so let's make something available like company fitness so that they can still be active during work” E10.
Participation
The employees believe that participation in using HealthySensor should always be voluntary: “There are always have people who don't want anything to do with it” E1. E8 argues that adoption of HealthySensor depends on whether it has a compelling character or not. Even if employees have previously indicated that they would like to participate; however, now have reservations they should be able to withdraw. “Can someone use an opt in or an opt out for this?” E9. By contrast, E2 argues that: “With health programs, I think they can indeed enforce it. We just put those chairs down in that office and you just get a different keyboard and you make sure you work with it”.
4.5.4. Innovation characteristics Costs
31
unreasonable” E7. Nevertheless, this only goes for the wearable as: “I wouldn't want to pay for a vibrating office chair” E7.Frequency
The employees have differing opinions on how often they should receive a reminder. Most employees believe that it should not be annoying, and they should have the option to ignore the alert. “It should not send a signal every half hour that you have to do something else” E7. E2 adds to that by arguing that he will quit if he receives notifications or reminders every five minutes. By contrast E4 argues the following: “First give a beep and let it get annoying. Then you do it again after a minute and you are going to let it come faster and faster until you really get agitated. There is a medical side, the system warns for a reason. That signal is because you are wrong, and you have to do something now”. Aesthetics
The employees identify aesthetics and comfort as important factors for using the devices. Next to measuring sitting position and duration, the chair should sit comfortably as well. E1 mentions that they are spoiled with 24-hour chairs and this chair should not provide less comfort. In addition, E3, E4 and E9 argue that the chair must be adjustable in multiple ways and believe in broader acceptance when the chair is automatically adjusted to an individual when sitting down “Ultimately, it comes down to that it must sit comfortably” E5. The results indicate that employees are concerned about the aesthetics of the wearable. They argue that wrist-worn devices should fit their unique preference and style. If a single device is offered and the aesthetics or brand do not suit the ‘taste’ of an individual, he/she will not wear it. People also have their own smart or analogue watches and are inclined to only wear it for a short period or not inclined to use a company issued device at all “For this to be successful, you will need to be flexible in your platforms. Not everyone wants to walk with a Fitbit or with an Apple watch” E3.
5. DISCUSSION
32
5.1. Comparison with current literature5.1.1. Attitude of employees towards SBET
Participants were more likely to have a positive attitude about introducing HealthySensor when they have experienced complaints from the musculoskeletal system or if they have an affinity with technology or data analysis. They believed it could be of assistance and help them in monitoring and improving health. Previous experience in health monitoring is a double-edged sword. Employees who have previously used software that locks their screens and forced them to take a break had a more sceptical view on its usefulness, while employees who only used fitness and health applications had a more embracing attitude. This is consistent with other research, as prior negative experiences tend to translate into a cynical attitude32,52,53.
5.1.2. Barriers and facilitators to employee engagement
The PEOU and PU, as reflected in the TAM-model34,35 were reinforced by the results as both barriers and facilitators. The findings showed that participants believe that HealthySensor should be easy to use and mentioned understandable language, having multiple platforms and advice presented in short and clear dialogue, which is also found by other authors54,55. Next to that participants perceive HealthySensor as useful when the advice is personalised to the user. The advice should be tailored to relate to topics that they perceive as fitting for their personal situation for it to be perceived as useful55,56. Furthermore, the findings indicate that adoption of HealthySensor is facilitated by employees’ self-efficacy and their knowledge and beliefs regarding healthy living. People are more likely to use and adhere to advice when they already believe in its merit57. In addition, participants indicated that aesthetics could be both a barrier and facilitator and is perceived an important factor in long term adherence29,57–59. When measuring or improving health does not interest them, they will stop using it quickly or refuse to use it at all60.
Employees argued that when provided with the choice to immediately follow advise or continue working on an urgent task. Most chose to keep working and argued that they could perform the action at a later moment in time. Receiving advice to take a break, while working on an urgent task could result in a user having the feeling that they have competing priorities; staying healthy and finishing work58. Sociodemographic factors, such as age, low educational level or limited technological skills, as well as eHealth literacy are also identified in literature as potential barrier to eHealth adoption61,62. Nevertheless, this was not found in this study. This is probably due to including participants who work with technology on a daily basis.
33
employer to encourage and facilitate usage of the intervention. Management support provides legitimacy to the intervention and motivates to continue58. Furthermore, the employees believed that the employer should cover the costs of the intervention. Not covering the costs raises a barrier for employees to use the tools. Furthermore, the employer should facilitate resources in the form of equipment and tools, as well as time to use the intervention. Users are not able to use the intervention when crucial information or resources are not available32.5.1.3. Barriers and facilitators to HCP adoption
The findings indicate that HCP perceive spending too much time on discussing results of the intervention as a barrier. The results from the intervention should be readable within a short period of time, so that they can focus on the treatment path during consult40. Furthermore, a lack of time and workload, as well as threatened clinical autonomy are identified in literature as barriers for physicians to use eHealth tools. Nevertheless, it was not mentioned by the participants in this study. Next to that, HCPs believed that results from the intervention will aid their work in determining treatment for an individual and functions as evidence to support the employees story38.
5.1.4. Peer contact and interaction
Peer contact and interaction with others is identified by the employees and HCPs as essential. Both employees and HCPs argued that users could motivate each other and be champions to involve others. Trusted co-workers or managers are a major influence on the users perception towards an intervention32. The HCPs argued that only providing SBET, without guidance and interaction with a doctor or others will not result in a change of behaviour or long-term engagement. This likely relates to algorithm aversion in which people prefer advise from a human over a computer system44. Only utilising eHealth interventions for self-assessment and improvement reduces the opportunity for users to receive support. As such, the HCPs argue for a mixed initiative in which self-management though HealthySensor is enhanced with support from peers or HCPs. This is in line with previous studies looking into self-management interventions55,63.
5.1.5. Design and technical concerns
34
5.2. Implications for theory and practiceStudies into facilitators and barriers of introducing eHealth have been conducted previously. Nevertheless, research introducing SBET in an office environment is scarce. The findings of this study contribute to theory by elucidating this underexposed subject. As such, this study identifies the views of desk-based employees on introducing SBET. In this study desk-based employees with different disciplines and job characteristics were interviewed. Therefore, the findings were yielded from differing contexts and provide an insight into a broad spectrum of desk-based occupations. Next to that, it takes the perspective of occupational physicians and experts into account. Their views on the HealthySensor concept, can be used as foundation for future research.
This study has several managerial implications. First, the results can be used by IMDI SPRINT to further develop the HealthySensor concept. They should consider adjusting the concept based on the identified preferences and needs of employees and HCPs. Next to that, they should address the design and technical issues raised by this research. Furthermore, the results provide change managers with footholds (facilitators and barriers) for devising an implementation plan when introducing SBET in their organisation. As such, the organisation and its employees can be prepared for the use of SBET.
5.3. Limitations and future research
In this research the views of 14 participants are presented on the introduction of SBET in the workplace and its implications for employees and HCPs. Nevertheless, there are several limitations related to this study. First, selection bias might have occurred as participants might have participated based on their interest in technology, data analysis or by snowballing. This was mitigated by including polar types of desk-based employees, participants with different job characteristics, to gain different perspectives on the subject65. Next to that, this study only included a small sample, as such results are not generalisable; however, that was not the aim of this study. Nevertheless, including more participants would yield a better substantiation of the results. As only 4 HCPs were included, a study examining specifically their view on the matter could expand on the results from this study.
