Behavioural and Societal Sciences Wassenaarseweg 56 2333 AL Leiden TNO report P.O. Box 2215 2301 CE Leiden The Netherlands TNO 2014 R10758 www.tno.nl How to Develop Personalized eHealth for Behavioural Change: Method & Example T +31 88 866 90 00 F +31 88 866 06 10 Date 30 May 2014 Author(s) Martin Laverman, MSc Prof. dr. Mark A. Neerincx Dr. Laurence L. Alpay Ing. Ton A.J.M. Rövekamp Prof. dr. Bertie J.H.M. Schonk Number of pages 115 (incl. appendices) Number of appendices 5 Sponsor Project name Periscope Project number 031.20999/01.02 All rights reserved. No part of this publication may be reproduced and/or published by print, photoprint, microfilm or any other means without the previous written consent of TNO. In case this report was drafted on instructions, the rights and obligations of contracting parties are subject to either the General Terms and Conditions for commissions to TNO, or the relevant agreement concluded between the contracting parties. Submitting the report for inspection to parties who have a direct interest is permitted. © 2014 TNO TNO report | TNO 2014 R10758 2 / 88 Preface In the current health care landscape, patients are more than ever expected to take an active approach in managing their condition. For patients with chronic conditions this means that they have to change their behaviour and develop new routines to deal with the day-to-day care of their disease. eHealth applications via websites or smartphone apps can offer essential support for these patients. Information and communication in these applications should match users’ needs and abilities. This report focuses on how such personalized eHealth applications to support behavioural change can be developed in an evidence-based and incremental approach. Following a situated Cognitive Engineering method, we will formulate the demands and requirements for personalized eHealth support and will identify socio-cognitive factors for personalizing information and communication. We will operationalize these theoretical insights into the design of a prototype personalized behavioural change support system and test this prototype with its design rationale in a feasibility study with prospective users of such systems. The research in this report has been carried out as part of the ZonMW (Dutch Organisation for Health Research and Development) programme Diseasemanagement for Chronic Illness, project Periscope (Personalized and Contextualized Information in Self-Management Systems for Chronically Ill Patients; projectnumber 300020001) from February 2010 to February 2014. The project is a collaboration between the Leiden University Medical Center and TNO, Netherlands. The authors would like to acknowledge the following people for their collaboration in this project. Dr. Paul van der Boog (LUMC) and Hannie Piels (www.mijnnierinzicht.nl) for their consultation and input at various stages in the project. Prof. Dov Te’eni (Tel Aviv University) for his insights in shaping the project and the design of personalized information. Dr. Akke Albada (NIVEL, Netherlands), dr. Yvonne Janssen (TNO, Netherlands) and dr. Yvo Sijpkens (Bronovo Hospital, The Hague, Netherlands) for their collaboration in the focus group interviews. Heleen van Tilburg (www.woordendiespreken.nl) and Frank Goethals and Rosie Paulissen (TNO, Netherlands) for their work on writing and designing the personalized information used in the feasibility study. TNO report | TNO 2014 R10758 3 / 88 Summary Self-managing your health and the complications and risks of chronic conditions need to be supported by care professionals and self -management support systems (SMSS). An SMSS is a computer system which helps patients to control the risks of his condition and maintain his health. Mostly, a change in life style and health behaviour is necessary, and it is exactly this change that only the person himself can achieve. In the Periscope project we have investigated which requirements an SMSS should meet to support users in changin g their life style with a personalized SMSS and how this support can be developed and tested in a systematic manner. We have chosen to investigate this in the domain of changing dietary habits, as this is an important factor to maintain health and prevent risks in many chronic conditions (e.g. diabetes and renal diseases). We know from the literature that form and content of information needs to be matched to rational (‘conscious’) as well as affective (‘unconscious’) information processing of the users of SMSS. This means that information needs to correspond to users’ personal situation, appreciations, motivation and cognitive capacities. The Periscope project has investigated how information in SMSS can take into account these personal factors, how this information should be designed and whether it is feasible to measure personal factors and determine the form of personalized information utulizing a prototype SMSS. The result of this research is a method that utilizes insights from previous research from the scientific literature, and user experiences and requirements to offer tailored information in SMSS. The developed method has been tested and evaluated by conducting an experiment with participants that have a chronic kidney condition who have to watch their diet very carefully using a prototype containing tailored messages. From this experiment we have concluded that interaction design patterns for tailored information can be applied in SMSS aimed at patients with chronic kidney conditions. One of the preconditions for implementing tailored information is that SMSS can reliably measure user characteristics and how users use the SMSS. Instruments that can measure this easy and reliable are being developed. The Periscope project has made an important step towards personalized information, but further research remains necessary into interaction design patterns for personalized information and the operationalization of theoretical insights into practical SMSS. TNO report | TNO 2014 R10758 | Final report 4 / 88 Contents Preface ...................................................................................................................... 2 Summary .................................................................................................................. 3 1 Introduction .............................................................................................................. 6 1.1 Behavioural change support in health care ............................................................... 6 1.2 Aim and Research Questions .................................................................................... 7 1.3 Methodology: situated Cognitive Engineering ........................................................... 7 1.4 Contents of chapters.................................................................................................. 9 2 Functional Model of Behavioural Change Support ............................................ 11 2.1 Need for evidence-based development of behavioural change support ................. 12 2.2 Method: Literature search ........................................................................................ 14 2.3 Critical success factors of support systems for behavioural change ....................... 14 2.4 Functional model of support systems for behavioural change ................................ 17 2.5 Conclusion & Discussion ......................................................................................... 21 3 User Perspectives on Behavioural Change Support ......................................... 23 3.1 Differences in perspectives on behavioural change support ................................... 24 3.2 Method: Focus group interview ............................................................................... 24 3.3 User perspectives on behavioural change support ................................................. 25 3.4 Implications of user perspectives for functional model ............................................ 29 3.5 Conclusion & Discussion ......................................................................................... 30 4 Socio-cognitive factors in Personalized Behavioural Change Support .......... 31 4.1 Tailored information for behavioural change support .............................................. 32 4.2 Method: Literature research .................................................................................... 33 4.3 Socio-cognitive factors for personalized behavioural change support .................... 33 4.4 Conclusion & Discussion ......................................................................................... 39 5 Designing Behavioural Change Support based on the Elaboration Likelihood Model ...................................................................................................................... 42 5.1 The Elaboration Likelihood Model as a basis for personalizing information ........... 43 5.2 Scenario and Use Cases ......................................................................................... 44 5.3 Claims and Requirements for personalized BCS based on ELM ............................ 46 5.4 Design Patterns for personalized BCS based on ELM ............................................ 49 5.5 Conclusion & Discussion ......................................................................................... 55 TNO report | TNO 2014 R10758 | Final report 5 / 88 6 Feasibility Study Behavioural Change Support Prototype for Chronic Kidney Disease ................................................................................................................... 56 6.1 Integrating Requirements into a prototype to study feasibility ................................. 57 6.2 Instantiating Strong Arguments and Affective Cues for reducing salt intake .......... 58 6.3 Method ..................................................................................................................... 59 7 Results of Feasibility Study Behavioural Change Support Prototype ............. 63 7.1 Introduction .............................................................................................................. 64 7.2 User characteristics ................................................................................................. 64 7.3 Main effects ............................................................................................................. 66 7.4 Interaction effects .................................................................................................... 71 7.5 Conclusion ............................................................................................................... 73 8 Conclusions & Recommendations ...................................................................... 75 8.1 Combined, dynamic and sensitive personalized information .................................. 75 8.2 Methodic development of BCS using sCE .............................................................. 76 8.3 Methodology in practice ........................................................................................... 76 8.4 Constraints of Feasibility Study ............................................................................... 77 8.5 Concluding remarks ................................................................................................. 77 9 References ............................................................................................................. 78 Appendices A Scenario Chronic Kidney Disease Self-Management B Use Cases C Strong Arguments and Affective Cues Stimuli D Questionnaires Feasibility Study E Distribution Plots TNO report | TNO 2014 R10758 | Final report 6 / 88 1 Introduction 1.1 Behavioural change support in health care The current health care landscape in The Netherlands is characterized by an aging population and rising prevalence of chronic conditions on the demand side of care, while health care providers are faced with austerity measures and the consequent need for greater efficiency (Dutch Council for Public Health and Health Care, 2010; Dutch Council for Publich Health and Health Care, 2010). One of the main approaches in future healthcare therefore is a greater emphasis on patients’ own responsibility and ability to take over care actions from providers, i.e. self- management (Wagner et al., 2001; Bodenheimer, Wagner, & Grumbach, 2002b). These chronic conditions are for an important part life style dependent, and require patients to change their behaviour and incorporate new routines in their life style. Thus, patients have a personal responsibility to take an active role in their care process, although this does not mean that patients are on their own in coping with their condition (Newman, Steed, & Mulligan, 2004). Several educational programs to help patients learn to self-manage and change their behaviour to cope with their 1 condition have been developed. eHealth solutions like websites and mobile phone applications are also being developed to suppor t patients in self - management by providing educational materials and guidance to change and maintain healthy life styles in accordance with their condition ( Portnoy, Scott - Sheldon, Johnson, & Carey , 2008 ; Sarasohn - Kahn, 2013 ) . We refer throughout the report to these systems as behavioural change support (BCS) and self - management support systems (SMSS). This report will focus on the role o f such eHealth systems to offer support for people who can benefit greatly from life style changes and developing tailored self - management support systems that accommodate personal prefe rences and skills of users . The use of SMSS to support patients to ch ange their behaviour has to date lead to moderate success. Two key challenges have arisen during the past years. The first challenge is to address the complexity of communication that results from using SMSS ( Te'eni, 2001 ) . Patients are much more on their own and have less direct feedback from professional caregivers when using such systems. Information that is communicated to patients by means of SMSS is aimed at increasing p atients’ knowledge of their condition and persuade them to change their life style to accommodate their disease. While information in SMSS, together with professional caregivers and patients’ social environment, can support patients to change their life style, patients have to make these changes themselves. It is therefore paramount to make sure that patients understand information and are convinced to take actions to stabilize or even improve their health. An important consideration in this regard is that there are differences between people in the way they process information and are triggered to change their behaviour. These cognitive differences need to be taken into account when communicating with patients via SMSS. 1 eHealth has been defined by Eysenbach (2001) as: “[…] the intersection of medical informatics, public health and business, referring to health services and information delivered or enhanced through the Internet and realted technologies. […] a state-of-mind, a way of thinking, an attitude, and a commitment for networked, global thinking, to improve health care […] by using information and communication technology.” TNO report | TNO 2014 R10758 | Final report 7 / 88 Tailoring information to the distinct ive circumstances and abilities of individual patients provides a means to account for these differences and thus improve the effectiveness of self -management support systems for behavioural change (Hawkins, Kreuter, Resnicow, Fishbein, & Dijkstra, 2008 ). However, tailoring to date has seen a focus on relatively broad concepts to determine which information needs to be communicated to which person. In this report we will advocate for a more dynamic and sensitive form of tailoring which takes into account people’s personal situation, appreciations, motivation and cognitive abilities to realize truly personalized information. In Chapters 4 and 8 we will further elaborate on these tailoring techniques. The second challenge concerns the development of eHealth based behavioural change support. The development of SMSS has been rather ad-hoc and aimed at supporting specific patient groups. A more methodological, user centered development would be preferable to provide evidence based and incremental development of such systems and that takes advantage of proven insights and solutions. By not focusing on specific clinical conditions, we aim to provide solutions that are not bound to one specific domain, but can be reused and adapted for different clinical conditions. This development should not be driven only from a medical perspective, but needs to take user requirements into account (Gustafson & Wyatt, 2004; Pal et al., 2013). Below we will present a methodology for evidence- based and incremental development of such support systems. This report will describe this methodology and show its applicability in the (e-)health care domain. 1.2 Aim and Research Questions The aim of this report is twofold. First, we will investigate which requirements SMSS must meet to support users to change their life style and how this personalized support can be designed and tested in a systematic manner. These requirements will be based on insights from previous research from the domains of self- management and behavioural change. Second, we will investigate whether the proposed methodology (see paragraph 1.3) offers a suitable framework for evidence-based and incremental design of SMSS functionalities and whether claims that are derived from previous research can be adapted to the self-management domain and can be validated in an online prototype. The research questions that follow from these aims are:  How can SMSS take the personal factors of users into account when communicating health information?  How should this personalized information look like?  Can the effectiveness of this personalized information be tested in online SMSS? 1.3 Methodology: situated Cognitive Engineering The research reported in this report has been guided by a situated Cognitive Engineering approach. Situated Cognitive Engineering (sCE) has been developed at TNO and Delft University of Technology by professor Mark Neerincx and TNO report | TNO 2014 R10758 | Final report 8 / 88 colleagues and aims to provide a theoretical and empirical driven user -centered design methodology for ICT based socio -cognitive support (Neerincx & Lindenberg, 2008; Neerincx, 2011). This methodology aims specifically at translating theoretical insights into the design of functionalities for support (i.e. a design dilemma), as opposed to for instance the Intervention Mapping approach (see e.g. Kok, Schaalma, Ruiter, Van Empelen, & Brug, 2004), which focuses on selecting and 2 testing the proper technology for support (i.e. a selection dilemma) . Situated Cognitive Engineering advocates an incremental and iterative development process in three phases. The methodology is i ncremental as it builds on proven theories and developed functionality can be reused in future development processes . It is iterative as it promotes insights that have been developed in one of the phases to be used to adjust and refine insights from the ot her phases. The sCE process (see Figure 1 . 1 ) starts with a Work Domain and Support Analysis (WDS) . The WDS incorporates a specification of the operational demands, i.e. which task(s) the system should support, soci o - cognitive theories that play a role in these tasks, and what technology is envisioned to be able to support these tasks. Based on the WDS , Core Func t ions (CF) that describe the specific functionalities that are needed to realize the operational demands n eed to be specified. Each CF should be accompanied by Claims on its operational effects, including possible positive and negative outcomes, to justify the incorporation of said CF, in favor of other functionality that could realize the same operational dem ands. Not always enough supporting evidence for the Claims can be derived from previous research, and in these instances, Claims should serve as hypotheses to be validated in the next phase of sCE (Refinement). To illustrate and organize the CF, Use Cases can be used. For clarity, these Use Cases need to refer to the CF and Claims they support, and can be incorporated in a scenario to present and discuss their rationale with stakeholders. Furthermore, they provide a means for incremental development: in fut ure systems where comparable Use Cases exist, previously developed CF can be reused. To describe the shape of CF (i.e. what does the CF look like) Interaction Design Patterns should be specified. Interaction Design Patterns offer a structured description o f the design of CF and include how it looks, in which context the design can be used and what rationale is behind the design. The CF and accompanying Use Cases, Interaction Design Patterns and Claims are part of the Requirements Baseline (RB) for the suppo rt system in development and serve as rationale and justification of these requirements. The third phase of sCE concerns the refinement and validation of the RB. This is an incremental and iterative process including review of the RB by prospective users and experts and prototype testing the functionality specified in the RB. By using prototypes, Claims concerning the operational effects of functionality can be validated in a real life situation. This report serves as an example of the application of sCE i n the health care domain. More extensive description s and examples of sCE from the space, defense and educational domains have been published, e.g. Neerincx (2011 ) , Neerincx et al. (2008 ) and Peeters, Van Den Bosch, Meyer, and Neerinc x (2012 ) . 2 The two approaches (sCE and IM) are as such complementary and a project aiming at combining both approaches is currently being developed at TNO, The Netherlands. TNO report | TNO 2014 R10758 | Final report 9 / 88 Figure 1.1 Situated Cognitive Engineering methodology. Italicized titles show how each step is Operational Socio-Cogenmibteidvdee d in this reEponrtv. isioned Work Domain & Demands Theories Technology Determinants for Support Analysis Functional Model of BCS Personalization Behavioural Change (Chapter 2 & 3) 1.4 Co(nCtheanpttse ro 4f )chapters Support Systems In this report, we focus on eHealth as a technology to support people who are at risk for or have a chronic illness who can benefit from a change in their life style. Design Patterns First, we will determine in general which functionalities are needed to offer tailored Strong Arguments and Affective Cues support to users in managing their health and life style. These operational demands (Chapter 5) for these systems are described in Chapters 2 and 3. Both chapters focus on the Use Cases Claims Patients Using Personalized BCS question of what aEnl aSbMorSatSio snh Loikueldlih loodk Mlikoed etol provide tailored self-management shape (Chapter 5) support. Chapter 2 bases(C dheamptaern 4d s& f 5ro)m a medical and scientific perspective on a illustrate literature study and ju psrtoifyvides a first instantiation of a reRqeuirqemueinrtse bmaselinet bsy organize providing a functional model of behavioural change support by SMSS. Chapter 3 Core refines these demands in a qualitative study with prospective usBears soef SlMinSeS. To esFtaubnlischt iwohnicsh personal factors can be utilized to offer personalized support with such system, we will address relevant socio-cognitive theoretical models from behavioural change and communication in Chapter 4. Chapters 2, 3 and 4 as such form the Work Domain and Support Analysis, and address the first research question posed in paragraph 1.2. Requirements Functional Model, DCehtearpmteinra n5t sw fiollr rPeeprsoornta olizna tihoen asnpde cification of an RB describing personalized Requirements for Pbeershoanvailiozeudr aBlC cSh (Cahnagpete sr u2,p 3p, o4r &t b5a) sed on the Elaboration Likelihood Model of Persuasion (ELM; Petty & Cacioppo, 1986). Claims that justify the use of different Core Functions for specific users to match their cognitive abilities and motivations, and use cases that describe the context in which these CFs are used, are provided Expert Prototype Refine & Refinement Review Testing Focus Group Interview Validate Feasibility Study Processes (Chapter 3) (Chapter 6 & 7) TNO report | TNO 2014 R10758 | Final report 10 / 88 in this chapter. Chapter 5 also describes the design patterns that shape these personalized Core Functions. This chapter hence addresses the second research question concerning how personalized information in SMSS should look like. We will continue with describing the Refineme nt Processes in the subsequent chapters. In Chapter 6 we describe a controlled experiment to show the feasibility of the personalized Core Functions that we have proposed. We have illustrated our RB with a scenario and several general use cases describing the context in which such personaized CF could be used, which are provided in Appendices A and B. Chapter 7 addresses a feasibility study to investigate the proposed personalized functionality. Both chapters concern the third research question of using an online SMSS to test the feasibility of the specified Requirements Baseline, and Chapter 6 offers additional insight in the research question concerning how personalized information should look like by instantiating Design Patterns into persuasive messages for SMSS. We will end in chapter 8 with the conclusions and recommendations for future development we can draw from our investigation into personalized information for BCS. To sum up, we will first formulate demands and requirements for tailored SMSS from socio-cognitive theories, medical literature and insights from prospective users of such systems. Second, we will investigate which socio-cognitive factors can be used to personalize information in SMSS. Third, we will show how to operationalize theoretical insights into practical requirements for personalized SMSS. The latter is done through systematically formulating and validating design patterns and claims concerning personalized persuasive information. We will conclude the report by summarizing what the key requirements for personalized BCS are, which socio- cognitive factors can be used to personalize information in BCS and how these socio-cognitive factors can be operationalized in personalized BCS.