Our computer program is comprised of six sections, plus an optional tutorial on how to use the program:
Getting Started describes the goals, importance and key components of advance care planning, provides an overview of the program and its various features, and explains how to navigate, complete exercises, save materials and print.
Choosing a Spokesperson reviews the concept of a surrogate decision-maker, including how to choose a spokesperson, designate someone as durable power of attorney and discuss decisions with this person. Through a question–answer format, users are then prompted to designate individuals to be their primary or alternate spokesperson(s).
Exploring Your Values asks users a series of questions about their personal values and goals regarding medical care, death and dying, and disability. This information becomes data for the decision aid that will help users make choices about medical treatment and generate a detailed advance directive for health care.
Your Medical Wishes helps users think about various medical interventions, and asks a series of questions whose answers contribute to the eventual detailed advance directive. This section uses video, photos, text and narratives to explain common health conditions (such as stroke, dementia, coma, and terminal illness) that can prevent a patient from communicating his or her preferences for medical treatments. There is a description of each condition and its consequences, what it is like to experience these medical conditions, and what treatment options are available. In similar fashion, this section describes medical interventions that commonly involve life and death decisions (i.e. CPR, mechanical ventilation, dialysis, tube feeding, and hospice/palliative care). Beyond the basic information provided for each condition and treatment, users may access additional information through an on-screen link. Subsequently, they are prompted to make a series of decisions about various conditions and treatments – which provides data used by the decision aid to individualize the eventual advance directive.
Putting It All Together is a summary section based on the information previously gathered. In this section, the program uses the decision aid to generate a printable document that articulates what the individual wants and who (if anyone) has been designated as a surrogate decision-maker. Included in this is one of the six general philosophical statements that an individual may accept, edit, or choose another to best reflect his or her general stance toward potentially life-sustaining medical interventions. Additionally, the generated advance directive contains the user’s specific wishes for particular medical conditions and treatments, including those states of being that would constitute (for that individual) an unacceptable quality of life. After confirming or revising these choices, the document may be saved, printed and distributed as a formal advance directive.
The Next Step reinforces the importance of communicating one’s wishes to loved ones, health-care professionals, and others who might be involved in medical decision making. Recommendations on how to initiate and sustain conversations on these issues are also provided.
Throughout the program, the user can access video clips of patient stories (or ‘testimonials’) for five individuals who describe their own experiences dealing with the particular topic being addressed (samples available for viewing at http://www.hmc.psu.edu/humanities/Advance%20Directives%20Project.htm). While the use of patient testimonials is controversial because of concerns about selection bias and unintended persuasion,32 we include them in our program primarily to portray varied perspectives, rather than to provide prescriptive influence about how decisions should be made. The differences in experience, age, gender, ethnicity and background are used to illustrate that advance care planning is relevant to a broad audience.
Users are encouraged to proceed through the program in the sequence provided, but are also free to navigate the program in any order they wish. That said, certain sections must be completed before the program can generate a final advance directive.
A description of our decision aid
The most innovative aspect of our program is the use of a decision aid to help individuals make choices about the medical treatment they would want if unable to speak for themselves. The decision aid is based on Multi-Attribute Utility Theory (MAUT), a type of decision analysis that systematically weighs competing objectives and can help translate a person’s values and goals into a medical plan of action for complex decisions.
Multi-Attribute Utility Theory’s key premises are (1) when choosing between alternatives, the best choice maximizes positive outcomes and minimizes negative ones33 and (2) a person’s choices can be accurately ranked by a mathematical formula that calculates his/her preferences for different aspects (or attributes) of choices. MAU models have been successfully applied to various health-care scenarios where there is no clear correct decision, such as mammography screening in younger women, family planning, flu vaccination and follow-up for abnormal Pap smears. But, to our knowledge, they have not yet been used for advance care planning. What makes MAUT particularly well-suited for advance care planning is its ability to account for individual attributes of a decision, weigh the relative importance of each attribute, and synthesize a person’s diverse values and desires into a single choice that optimizes their preferences. We have integrated a multi-attribute utility model into our computer program as a way to help individuals reflect on decisions, prioritize their values, and develop a rational plan that is consistent with their wishes.
Development of the MAU model
The MAU model we use in our program follows a six-step approach:34 (1) identify decisions that need to be made; (2) identify important attributes of the decisions to be made; (3) have individuals assign important weights to each attribute; (4) have individuals rate the decision components using the attribute matrix previously developed; (5) calculate an overall preferences score for each decision, and (6) return results back to participants for revisions and/or confirmation.
To identify and refine the ‘attributes’ (i.e. factors) influencing individuals’ decisions whether to receive life-saving medical treatment, we first reviewed the published literature on qualitative assessments of individuals’ preferences for end-of-life care. After identifying a preliminary list of attributes, we next conducted four individual interviews and four focus groups to refine this list and to test it for face and content validity. We enrolled 23 individuals, including nine from a community geriatrics centre, seven from an urban community center with a predominantly minority population, and seven from cancer support groups at an academic medical center.
Using accepted focus group methodology,35 we asked participants to identify factors that would influence their decisions whether or not to receive life-saving medical treatments, and then facilitated a semi-structured discussion to identify common themes and patterns. All interviews and focus groups were audiotaped and transcribed, and two investigators independently reviewed the transcripts, identified themes, compared results, and reached consensus about categories. Participants were eager to discuss end-of-life planning, often reporting the lack of opportunity to do so with health-care providers. Content analysis of focus group transcripts demonstrated a strong consensus that the following attributes were most important in deciding whether to receive medical treatment when unable to speak for oneself. As such, they were integrated into the MAU model of our program, where they are rank ordered by users, and then assigned numerical scores so as to weight their priority.
The physical symptoms associated with the condition.
The effect of the condition on one’s mental functioning.
The effect of the condition on one’s independence.
The prognosis of the condition.
Whether the condition would make one a burden to others.
The burden of the medical treatment.
Whether the treatment would make one a burden to others.
The list of attributes are integrated into the program to help the individual make decisions about the use of particular interventions under a variety of clinical circumstances. For example, users are asked to imagine suffering a moderate/severe stroke that would improve during the next year and to consider whether they would want an intervention such as mechanical ventilation for that condition. By assessing how the condition and intervention rate for various attributes (e.g. whether mechanical ventilation for a stroke is burdensome to oneself or others), the program (via the MAU model) is able to prioritize the individual wishes for a variety of condition-treatment combinations.
The program also helps individuals articulate what counts as ‘poor quality of life’ by asking how acceptable various states of dysfunction would be to them. Users articulate whether particular situations would be ‘acceptable’, ‘acceptable but difficult’, ‘worth living but just barely’, or ‘not worth living’ (with a fifth option being ‘not sure’). Five physical aspects are explored (pain, discomfort, incontinence, and moderate or severe immobility), three mental aspects (communication, confusion, decision making), and four social aspects (independent living, relationship, financial burden, and burden to family). Those ‘states of being’ judged to be either ‘not worth living’ or ‘worth living but just barely’ are subsequently listed in the advance directive as indications of what ‘poor quality of life’ means for that individual.
In the final part of the Putting It All Together section, individuals are given the opportunity to review and revise their choices for various conditions and treatments. Additionally, they are asked whether they wish to revise their choice of spokesperson(s); whether they want their spokesperson or their written advance directive to take precedence in the event of a conflict; whether they would want to participate in clinical research; whether they wish to be an organ donor and whether there are specific personal or spiritual wishes they would like to express. When users are satisfied that they have accurately expressed their wishes, they can print an advance directive document that can be signed, witnessed, and given to physicians and loved ones.
Preliminary testing of the program has shown that individuals find it easy to use and they perceive it to be accurate at representing their values and preferences. In a pilot evaluation with 50 adult volunteers recruited from an internal medicine outpatient practice in central Pennsylvania (mean age 52 years, 68% female, 68% college graduates), users spent an average of 106 min completing the program, and they indicated that this duration of time was not burdensome. Users were very satisfied with the program overall (mean satisfaction = 8.5, where 1 = not at all satisfied and 10 = extremely satisfied), and particularly with how it improved their knowledge and helped them make decisions (mean 4.2, where 1 = very dissatisfied and 5 = very satisfied).
In another pilot study, 34 individuals with cancer were recruited from clinics at Penn State Hershey Medical Center (mean age 57 years, 71% female, 53% with breast or lung cancer). Satisfaction was also very high in this group (mean = 8.5, where 1 = not at all satisfied and 10 = extremely satisfied). Furthermore, users indicated that the program was highly accurate at representing their wishes; prior to making edits to the computer-generated advance directive, mean accuracy was 5.5 (1 = not at all accurate, 7 = very accurate), and this increased to 6.5 post-editing (P < 0.001). And, as we expected, users had no change in levels of hopefulness, hopelessness, or anxiety following the intervention.
Some anticipated critiques and our response
While we believe our interactive computer program offers numerous advantages over standard advance directive forms, no program is perfect, and we anticipate several potential critiques.
(1) A computer program is too complex for those who are old, sick, or poorly educated.
Though computers cannot overcome all the barriers to advance care planning (nor should they be seen as a replacement for medical professionals), they do have the potential to help individuals be better prepared for end-of-life health-care decisions. Computers have been shown to be widely accepted by people regardless of socioeconomic status, educational background, or age.36 Even in older adults with no prior computer experience and individuals with low literacy skills, computer-based education has outperformed traditional text-based education,37 and has been shown to increase self-esteem, as well as self-perceived productivity38 and autonomy.39
The use of computers in health care is widely accepted, and numerous studies have confirmed their effectiveness and acceptability for patient education,40 social support,41 taking medical histories,42 improving surgical management,43 promoting patient preferences44 and adherence to medication,45 and even providing psychological counselling.46 Our own work has demonstrated that computers can increase knowledge about breast cancer genetic testing and help people make difficult, value-laden choices.47,48 Add to this the ability of computers to compensate for sensory deficits such as impaired vision or hearing, and the potential impact of a well-designed computer program for advance care planning is enormous.
(2) People are not able to make meaningful and/or reliable decisions about medical conditions with which they have had little or no experience. Future forecasting is notoriously inaccurate, and people often overestimate the negative aspects of illness and underestimate their ability to cope with physical challenges.49
The problem with inaccurate future forecasting is a serious one, but not specific to computer-based advance care planning. While people may not be able to predict with absolute confidence what they would want in the future, our program is designed to inform and educate, using vignettes, personal testimonials, didactic tutorials, and values-clarifying questions. Even if an individual fails to accurately predict what they will eventually want, we believe our program gives them a viable opportunity to learn about and reflect on potential future scenarios. Furthermore, the alternative to planning and predicting is to NOT plan or predict, which leaves decisions in the hands of people who are even less qualified to know what a person would want. Thus, in our assessment, it is better to try and sometimes be wrong than to not try at all.
(3) Given people's difficulties with predicting what they would want for themselves, we should not rely on surrogate decision-makers to make good, representative decisions on behalf of others.7
The literature on surrogate decision making reveals that surrogates tend to be poor predictors of patients’ wishes, incorrectly predicting their preferences one-third of the time.15 In an effort to overcome this problem, our program emphasizes the importance of discussing one’s wishes with both spokesperson(s) and health-care providers. In addition to specific advice about how to initiate and carry out such conversations, the program also lends itself to being used jointly, and models the conversational approach that we believe can improve the quality and accuracy of surrogate decision making.
The bottom line is that medical decisions at the end of life are not optional; they must be made one way or another. Hence, the question is what is the best way to help those who must make these often-difficult decisions and can our computer-based decision aid can do a better job than standard advance directives? We contend that individuals who learn about the nature and implications of common end-of-life conditions and interventions will be better prepared to make decisions that are consistent with their wishes. For this reason, education is a major focus for our program: providing information, opportunity for reflection, and encouraging people to think through the implications of their choices. That people might change their minds or poorly predict their future wishes speaks to a limitation of all anticipatory decision making. In the case of surrogates, who often must make end-of-life decisions for patients, we contend that they will be better prepared to accurately represent patients’ wishes to the extent that such decisions are anticipated, reflected on and discussed.