A Holistic Approach to Enhance the Doctor-Patient Relationship for Diabetes Using Social Networking, Personalized Alerts, Reminders, and Recommendations William WL Yip Luz M. Quiroga University of Hawaii, Manoa University of Hawaii, Manoa 1680 East-West Road, 1680 East-West Road, POST 314B, Honolulu, HI 96822, USA Honolulu, HI 96822, USA 1-808-956-9988 1-808-956-9988 wyip@hawaii.edu lquiroga@hawaii.edu Abstract gap is attributed by the fact that most health care providers resort to the traditional model of compliance and adherence to treat This paper describes an ongoing project that proposes the chronic illnesses like diabetes [2]. This model, which was based conceptual design of a decision-support system (DSS) based on on a health care system that provided the majority of its treatment patient modeling that enhances the communication and for acute illnesses [1], can have potential damaging effects on the relationship among health care providers and patients with provider-patient relationship. Instead, the empowerment diabetes. This project attempts to answer the following two approach emerged in the early 1990's as a new model to promote research questions: 1) What are the challenges in the current equal partnership among providers and diabetic patients [7, 9, 12, relationship between a diabetic patient and his/her health care 17]. A DSS based on patient modeling can potentially facilitate providers? 2) Can a DSS based on providing motivation support this new approach. Effective communication among health care through social networking, personalized alerts, reminders, and providers and patients can be facilitated by a DSS that: recommendations improve objective and subjective factors that affect the overall health outcome of a diabetic patient? For patient: • Provides motivational support through social networking Categories and Subject Descriptors sites (SNSs). H.4.2 [Information Systems Applications]: Types of Systems – Decision support (e.g. MIS) • Provides alerts and reminders to motivate patient to comply with lifestyle-changing activities. General Terms • Provides personalized recommendations of trusted health- Design, Experimentation, Human Factors related information based on individual patient’s situation. For health care provider: Keywords • Provides personalized alerts and reminders when his/her Compliance, Diabetes, Empowerment, Information Filtering, patient’s physiological parameters (e.g. blood glucose level) Personalization, Social Networking are out of range. • Provides personalized recommendations of treatment options 1. Introduction based on evidence-based guidelines. This paper describes an ongoing project that proposes the Situation of each patient is unique. In [13], the study showed the conceptual design of a decision-support system (DSS) based on importance of context in users’ relevance feedback in information patient modeling that enhances the communication and filtering (IF) systems for delivery of personalized consumer health relationship among health care providers (i.e. physicians and information. The study identified non-topical characteristics such nurses) and patients with diabetes. With more than 23 million as lifestyle, domain expertise, credibility of information sources, Americans suffering from diabetes [3], health care providers and and comprehensibility. To increase patients’ motivation to change researchers have devised ways to improve diabetic patients’ behavior, providing the right information to the right patient at the overall health outcome as well as to reduce expensive acute right moment is crucial. To achieve this, a holistic model of the episodes as a result of non-compliant lifestyle activities [14]. In patients is required. In the health care domain, many spite of these efforts, there remains a gap in the communication opportunities exist for profiling patients. A holistic model may channel among health care providers and diabetic patients. This include: • Electronic Medical Record (EMR) • Vital signs and physiological parameters collected from outpatients settings (e.g. blood pressure, blood glucose level) • Quality-of-Life (QOL) issues (e.g. food intake, exercise) • Web browsing behavior that includes health-related diabetic patients to access information relevant to their individual websites, social networking sites (SNSs), and patient support situations more efficiently. Furthermore, the agenda service groups allows patients to generate a list of 5 questions that they can bring to their health care providers during their office visits. Although This information can be fed to an agent-based DSS which in turn this research shows the face validity of such information tailoring provides alerts, reminders, and recommendations to both health system for diabetic patients, it falls short of being a care providers and patients. comprehensive approach for both diabetic patients and their 2. Conceptual Framework health care providers. Such approach can help bridge the Why do diabetic patients remain non-compliant to lifestyle- communication gap and provide an environment of equal changing activities in spite of the physical, psychological, and partnership among all stakeholders. Research has shown that a financial burdens that diabetes place on them? The problem lies decision-support system can help health care providers follow in the application of the compliance and adherence model in clinical guidelines, which eventually leads to improved care [15]. diabetic care. In chronic diseases like diabetes, this model places Motivation is a key component in successful management of the patients in a submissive position obeying authoritative care diabetes. Self-determination theory distinguishes between providers [8]. Health care providers often feel frustrated with autonomous and controlled behaviors [18, 19]. Patients are their patients’ non-compliant activities. Vice versa, patients feel autonomously motivated when their desire to change behaviors frustrated with their lack of knowledge and understanding of the comes from within themselves; while behaviors are controlled disease as well as blames from their care providers for their non- when patients are pressured from external forces to change their compliant activities. What is needed is an approach that: 1) daily activities. Two separate studies [18, 19] showed that redefines the roles and responsibilities of both patients and care patients’ autonomous motivation is strongly correlated to their providers; 2) create a relationship that promotes collaboration and perception of their providers’ autonomous support. It is partnership [5]. Considerable amount of research has been done autonomous motivation that leads to patients’ competence in to facilitate this approach. Since the early 1990’s, there has been making lifestyle changes, and is therefore an important factor to a push for patient empowerment that gives control to both patients be considered in reducing the communication gap among diabetic and care providers [7, 9, 12, 17]. In [5], the authors even patients and health care providers. downplayed the importance of compliance, claiming that compliance becomes irrelevant if patients are “viewed as Social networking sites (SNSs), mostly in terms of support groups collaborators who establish their own goals”. In a community- around health issues, has a long tradition, starting with first based diabetes self-management education program [6], the study generations of social tools of the 1980’s exemplified by the encouraged patients to find their own solutions that fit their “Well” community in Rheingold’s book “The Virtual psychological and physical needs. In a study conducted with 85 Community” to activities such as Sermo (http://www.sermo.com/) type-2 diabetic patients [12], individuality was identified as one of - social networking for licensed physicians, NurseConnect the five issues that are pivotal to effective management of the (http://www.nurseconnect.com/),Nurses’ Lounge disease. It is a patient-centered approach where information (http://www.nurseslounge.com), and specific groups (by illness, delivered to patients is based on their individual needs and treatment, therapies, etc). More recently, progress has been made concerns. in Second Life (http://secondlife.com) with islands such as Health Info Island, Karuna (AIDS), Virtual Ability Island (disabilities), Diabetic patients need to make informed decisions in order to and Rachelville (parents of terminally-ill kids). Virtual events are manage their disease effectively. Informed decisions are based on held to promote “social engagement” such as the recent Helen information provided to patients pertaining to their individual Keller Day, organized by EASI: Equal Access to Software and needs and circumstances. Personalized recommendations are Information (http://easi.cc), as part of its commitment with provided to diabetic patients in a health information tailoring students and professionals with disabilities to have the same right system called Violet Technology (VT) in [9]. VT is a web portal to access information technology as everyone else. that performs information filtering and prioritization based on patients’ profiles in Diabetes Information Profile (DIP). There are In a pilot study [11], five participants were interviewed on their 5 components in DIP: perceptions (both positive and negative aspects) on SNSs and which properties of SNSs can facilitate online support and • Diabetes-related situation: current lifestyle, diabetes adherence to health-related regimens. The study found that SNSs education exposure, self blood glucose tests, medications. were most instrumental in providing emotional (e.g. encouragement from a friend) and informational (e.g. a tip to • History of information browsing perform a task) support. Furthermore, users of SNSs tend to build • Patient information preference and strengthen existing relationships among family members and friends rather than to meet new friends. This preliminary result • History of quizzing suggests that SNSs can potentially have a positive effect for • History of agenda generation diabetic patients who rely on their close ones for motivation. In the following section, a solution incorporating personalized alerts, The presentation of the information is adapted based on a two- reminders, and recommendations, as well as social networking step process. First, information is filtered using a series of rules features will be proposed. (e.g. removing female issues for male patients). Then, information is ordered by its significance based on priority assigned to each information item. The patient-modeling approach of VT allows 3. Solution Patients’ web browsing behaviors provide a clue on what their This section provides a brief description of each component of the information needs are. For instance, a patient who often searches DSS that this project proposes. Figure 1 below illustrates a for information about a particular drug indicates that he may be conceptual diagram of an agent-based DSS that provides prescribed with the drug and in need for additional information personalized alerts, reminders, and recommendations, as well as (e.g. recent findings on side effects). A DSS can provide motivations through social networking. personalized recommendation of health-related information from trusted source (e.g. National Library of Medicine, MedlinePlus). From the health care provider’s perspective, patients’ clinical data includes demographic information about the patient, clinical test results, history of drug prescriptions, history of vital signs and physiological parameters, and etc. This, typically in the form of Electronic Medical Record (EMR), is an enormous source of information that provides many opportunities for decision-support services based on evidence-based guidelines. Integrating evidence-based guidelines with EMR can help reduce practice variability and improves the overall quality of care for patients. The last component of the DSS is an online social network for Figure 1 Conceptual Design of a design-support system for both diabetic patients and health care providers. Diabetic diabetic patients and health care providers patients, health care providers, family members, and friends participate in a common medium to provide emotional and Although they provide similar functions, it is important to point information support for each other. Members can write messages out the minor differences between alerts, reminders, and in public (e.g. a “wall” on Facebook) as well as private areas. recommendations. The following scenario illustrates the Patients can also post updates on their health status. In addition, differences: games and puzzles can be used to educate members, increase John is a 56-year-old man who was diagnosed with diabetes participation, and keep members interested over a longer duration. 5 years ago. He monitors his blood glucose level daily using a blood glucose meter provided by his primary care 4. Research Methodology physician. He receives an alert from the meter when his This ongoing project proposes the following research blood glucose level is 10% above his acceptable range. He methodology, which is broken down into 3 phases. has an office visit with his dietician every 6 months. He creates a reminder on Google Calendar to remind himself of Phase 1 of the project will focus on the relationship among the appointments. Lastly, John’s brother was recently diabetic patients and their respective care providers. The research diagnosed with diabetes so John sent an online article from question of this phase is: WebMD about recommendations on how to perform foot What are the challenges in the current relationship between care on a periodic basis. a diabetic patient and his/her health care providers? Why From the patients’ perspective, there are 3 primary data sources are diabetic patients non-compliant to lifestyle-changing where profiling information can be collected: 1) vital signs and activities? physiological parameters; 2) QOL issues; and 3) web browsing To answer this research question, in-depth interview sessions will behaviors. Vital signs and physiological parameters are be conducted with diabetic patient educators and coordinators measurements that are collected periodically by patients from various health institutions. Patient educators act as themselves in remote settings. Examples are blood glucose level intermediaries between patients and care providers who can and blood pressure. QOL issues are qualitative indicators of how provide their unbiased opinions. In a way, they are “human well patients are managing their disease. QOL issues may be the agents” that perform similar functions that a potential DSS could amount of exercise that a patient is performing daily, or the do. The focus of the interview questions will center on the lifestyle preferences of the patients (e.g. smoker, preference to existing communication means (or lack thereof) among diabetic alternative treatment, their personal goal, plans, strategies, success patients and health care providers. and impediments regarding their management of the disease). Together, quantitative measurements of vital signs and Phase 2 of this project involves the conceptual design of the social physiological parameters and qualitative indicators of QOL issues networking component of a DSS. The specifications and design form a unique model of each patient. In an exploratory study of this component will be based on interview results from Phase [10], they designed the CHAP (Continuous Health Awareness 1. In Phase 2, the prototyped social networking component will Program) system that engage patients to reflect on their be integrated with an existing DSS called Comprehensive Disease breakdown activities and to build correlations between these Management Program (CDMP) (http://www.estenda.com), which activities and the collected data on the patients’ blood glucose is based on the Chronic Care Model [16]. CDMP, currently values. Based on the collected data on an individual’s vital signs, operational in more than 70 clinics in the Indian Health Service, physiological parameters, and QOL issues, a DSS can provide allows patients to upload their physiological parameters (e.g. personalized alerts, reminders, and recommendations. blood glucose) and images. This data is then combined with laboratory results and other patient records to provide decision- support services to patients and health care providers. The social networking component implemented in this phase will build upon [5] Funnell, M. M., & Anderson, R. M. (2000). The problem the existing patient profile in CDMP. with compliance in diabetes. The Journal of the American The last phase of this project is to evaluate the effectiveness of the Medical Association, 284(13), 1709. social networking component of a DSS implemented in Phase 2. [6] Funnell, M. M., & Anderson, R. M. (2002). Working toward The research question of this phase is: the next generation of diabetes self-management education. Can a DSS based on providing motivation support through American Journal of Preventive Medicine, 22(4), 3-5. social networking improve objective and subjective factors [7] Funnell, M. M., & Anderson, R. M. (2003). Patient that affect the overall health outcome of diabetic patients? empowerment: a look back, a look ahead. Diabetes Educator, 29(3), 454-464. The independent variable of this experimental study is the existence of a social networking component in health care setting; [8] Lutfey, K. E., & Wishner, W. J. (1999). Beyond while the dependent variable is the objective and subjective "compliance" is "adherence". Diabetes Care, 22(4), 635-639. factors that affect a diabetic patient’s overall health outcome. The [9] Ma, C., Warren, J., Phillips, P., & Stanek, J. (2006). control group is a group of diabetic patients and health care Empowering patients with essential information and providers who will not be provided with the social networking communication support in the context of diabetes. component; while the experimental group will evaluate the social International Journal of Medical Informatics, 75(8), 577-596. networking component implemented in Phase 2. About 50 [10] Mamykina, L., & Mynatt, E. D. (2007). Investigating and participants will be chosen from existing users of the CDMP into supporting health management practices of individuals with the control and experimental groups, respectively. diabetes. In Proceedings of the 1st ACM SIGMOBILE Objective factors are facts that can be measured quantitatively international workshop on Systems and networking support (e.g. patient compliance to measuring blood glucose level, number for healthcare and assisted living environments. San Juan, of times a patient is admitted to a hospital because of Puerto Rico: ACM New York. complications). On the other hand, subjective factors (e.g. [11] Olsen, E., & Kraft, P. (2009). ePsychology: a pilot study on motivation to comply with lifestyle-changing activities, patients’ how to enhance social support and adherence in digital perception of the health care system) are more difficult to measure interventions by characteristics from social networking sites. quantitatively. In this phase, both qualitative and quantitative data In Proceedings of the 4th International Conference on will be collected. Objective factors will be measured through a Persuasive Technology (pp. 1-6). Claremont, California: self-reported questionnaire comprised of 6 compliance ACM New York. components: exercise, hypoglycaemia, foot care, diet, home monitoring, and drug [4]. Statistical analysis will be performed to [12] Pooley, C. G., Gerrard, C., Hollis, S., Morton, S., & Astbury, measure the effects of the existence of the social networking J. (2001). 'Oh it's a wonderful practice... you can talk to component. Subjective factors will be evaluated through an them': a qualitative study of patients' and health interview session with a subset of the experimental group. The professionals' views on the management of type 2 diabetes. focus of the interview session is to determine whether social Health & Social Care in the Community, 9(5), 318-326. networking has an effect on diabetic patients’ perception on their [13] Quiroga, L. M., & Mostafa, J. (2002). An experiment in health care providers. Aspects of the perception of their health building profiles in information filtering: the role of context care providers include trust, availability, satisfaction, optimism, of user relevance feedback. Information Processing and and etc. These factors correspond to the autonomous support that Management, 38(5), 671-694. self-determination theory identifies and have a major effect on [14] Steinbrook, R. (2006). Facing the diabetes epidemic: patients’ ability to comply to lifestyle-changing activities. Mandatory reporting of glycosylated hemoglobin values in New York City. New England Journal of Medicine, 354, 5. References 545-548. [1] Anderson, R. M., & Funnell, M. M. (2000). Compliance and [15] Vashitz, G., Meyer, J., Parmet, Y., Peleg, R., Goldfarb, D., adherence are dysfunctional concepts in diabetes care. The Porath, A., et al. (2009). Defining and measuring Diabetes Educator, 26(4), 597. physicians’ responses to clinical reminders. Journal of [2] Anderson, R. M., & Funnell, M. M. (2005). Patient Biomedical Informatics, 42(2), 317-326. empowerment: reflections on the challenge of fostering the [16] Wagner, E. H. (1998). Chronic disease management: what adoption of a new paradigm. Patient Education and will it take to improve care for chronic illness? Effective Counseling, 57(2), 153-157. Clinical Practice, 1, 2-4. [3] Centers for Disease Control and Prevention. (2008). National [17] Weiss, M. A. (2006). Empowerment: a patient's perspective. diabetes fact sheet: general information and national Diabetes Spectrum, 19(2), 116-118. estimates on diabetes in the United States, 2007. Atlanta, Georgia: U.S. Department of Health and Human Services, [18] Williams, G. C., Freedman, Z. R., & Deci, E. L. (1998). Centers for Disease Control and Prevention. Supporting autonomy to motivate patients with diabetes for glucose control. Diabetes Care, 21(10), 1644-1651. [4] Chan, Y. M., & Molassiotis, A. (1999). The relationship between diabetes knowledge and compliance among Chinese [19] Williams, G. C., Rodin, G. C., Ryan, R. M., Grolnick, W. S., with non-insulin dependent diabetes mellitus in Hong Kong. & Deci, E. L. (1998). Autonomous regulation and long-term Journal of Advanced Nursing, 30(2), 431-438. medication adherence in adult outpatients. Health Psychology, 17, 269-276.