“Our culture is embedded with a strong belief that more is better and that physicians know best.” John Wennberg, MD
Clinical practice varies across regions of the United States. The variation is not explained by patient illness or preferences; the supply (or oversupply) of medical treatments impacts quality (or failure) of our health system. The United States has focused on medical errors and associated performance measures to reduce practice variation. Although important, it is misleading to solely measure the quality of how medicine is administered without also considering if medical treatments should have been administered in the first place. In the book, “Overtreated: Why Too Much Medicine is Making Us Sicker and Poorer,” Shannon Brownlee tells the non-obvious story of how more medical care often does not result in better outcomes. As a result, our health system compared to other developed countries is forty percent more expensive, yet it is in the bottom quartile for most quality indicators (Brook et al., 2000; Institute of Medicine, 2001).
In this article, I first discuss the exciting interdisciplinary field of medical decision making. Second, I describe how improving the medical decision making process (driven partly by empowering patients) can shift geographic patterns associated with over-treatment. Finally, I argue that the US healthcare system could be improved if patients and physicians use internet-based decision aids to guide the complex and often difficult decisions involving medical treatments.
Doing the Right Things Right
Clinical Evidence BMJ evaluated 2,500 medical treatments and found that only half of these were governed by clear evidence-based guidelines (How Much Do We Know?, 2006). Acknowledging this finding, O'Connor et al. (2007) defined "effective" and "preference-sensitive" treatments. Both have some scientific knowledge about outcomes, yet for effective treatments there is enough information that benefits outweigh harms. In contrast, for preference-sensitive treatments, we are less certain that potential harms are justified. Given the evidence that medical counseling is often inadequate, and that the use of preference-sensitive treatments can vary two to fivefold across practice regions, the authors posit that it is important to improve the medical decision process. Wennberg (2004) provides a good example by showing that the rate of exposure to coronary bypass surgery and the case fatality rate of bypass surgery patients is about the same. This be explained by the finding that a substantial percentage (between 0-25%) of CABG procedures are inappropriately performed in the first place (Summaries for Patients, 2001). If patients received medical treatments only when needed, it would make sense to only look at the quality of treatments. However, with varying appropriateness of treatments being performed, we need to improve the medical decision making process. Doing a treatment effectively is not enough; clinicians need to do the right things right (Chassin et al., 1987; Blumenthal, 2004; Wennberg, 2004).
The current focus of the healthcare quality movement is to measure medical errors. It is necessary, however, to measure both how well treatments are administered, and the quality of the decision making process (Sirovich et al., 2008; Wennberg, 2004). Although there are many possible tools to improve medical decision making, the focus here is on internet-based decision aids. Once patients think more clearly about their goals, values, and risk/benefits preferences, they may be able to help their doctors make better decisions for themselves and society.
Medical Decision Making
An excellent book titled, “Medical Decision Making: A Physicians Guide” bridges the gap between decades of decision science research and the actual use of effective decision methods among clinicians and patients (Schwartz and Bergus, 2008). The authors first discuss the importance of helping patients define their goals with respect to their health and medical treatments. Patients have diverse demographic traits and social situations, and these will shape how medical treatments can fulfill their near and distant objectives. Next, there is a large volume of research about how people rate their current health, and using these methodologies, it is possible to have a patient “trade-off” the length and quality of their life. For example, it is possible to estimate “quality-adjusted life expectancy” after obtaining information about health states, life expectancy of the patient, and a discounting factor for future health. This information can be used to construct decision models that can then be used for specific cases in which patients can decide if an extended life is preferable to treatment-induced disabilities. Although some people without question seek a longer life, others are not willing to give up a standard quality of life, and will choose more or less aggressive treatment options. These methods allow patients' values to be an important part of the medical decision making process.
The third section of the book stresses that physicians and patients need to have the tools to embrace uncertainty. The authors start by introducing the concept of risk. Risk, unlike uncertainty, allows one to assign a probability of the likely outcome based on data from a large sample of observations. For example, it is possible to assign a probability of death for a population of patients if we have quality outcome data for a population with similar age and clinical characteristics . Of course, assigning an individual patient an exact probability of the outcome is impossible; we have to accept that some individuals are different than the population in unknown ways, and thus predicting at the individual level is problematic. Information about risk at the population level, however, can still be very useful to patients if it is presented in a clear way. Gigerenzer (2002) provides a good example. He finds that people are better able to understand raw frequencies in contrast to probabilities. Thus, when presenting people with information about the risk of harm that they might experience from a medical procedure, it is better to show them graphical representations of numerators and denominators rather than rates and associated probabilities. Finally, the book provides an overview of the concept of expected value in which patients can evaluate expected events by summing up the value of an outcome multiplied by its probability of occurrence.
Although there is much more to learn about the medical decision making process, researchers have generated an impressive amount of knowledge that could help patients and physicians make better decisions. Much of this science is applied in nature and very applicable to the design of effective decision aids.
Patient Decision Aids Are Effective
High quality patient decision aids are based on knowledge from medical decision science. The tools help patients document their values and goals, and then help them to evaluate associated risks and benefits of their treatment options. To construct quality decision aids, evidence-based clinical guidelines and other clinical knowledge has to be summarized for each medical condition or treatment of interest. These should then be tested with a pilot group of patients to ensure that the tool is easily understood by patients. Since decision aids are often involve a tree structure or a set of questions to evaluate, they can be administered in-person, pencil-and-paper or through an internet application. Given the interest in provider websites and personal health records, it is likely that web-based applications will become an increasingly effective way to administer decision aids.
O'Connor et al. (2007) state that decision aids lead to informed judgments by providing patients with personalized information about the tradeoffs of their various treatment options. Following decision theory, they write that decision aids: (1) state treatment options, outcomes, and probabilities, (2) help patients' evaluate which outcomes are the most important to them and (3) provide clear steps for patients to make choices and then communicate these ideas to their physician.
Key points from O'Connor (2007, p.176) have been copied below to highlight the likely benefits of decision aids.
• “Patient decision aids prepare patients to discuss "grey zone" treatment options and to clarify
which benefits and harms matter most to them”
• “They differ from educational aids by not only providing option information but also tailoring it to the patient's clinical risk profile and guiding patients to express their personal values”
• “They promote informed values-based decisions and improve patient participation and comfort
levels in decision-making”
• “They prevent the overuse of "grey zone" options that informed patients do not value (e.g.,
aggressive discretionary surgery such as hysterectomy, prostatectomy and back surgery)”
• “They will be used routinely at the point of clinical care only if they are made easy for clinicians to use and become something patients expect as part of high-quality care”
Is there any strong scientific evidence that decision aids work? A recent meta-analysis published by the Cochrane Collaboration evaluates fifty-five studies of decision aids that used the gold standard randomized control trail (O'Connor et al., 2009). The authors found that decision aids were better than treatment as usual options. Specifically they found that patients using decision aids had: (1) more knowledge about the treatment (2) less decisional conflict and (3) were less likely to be passive in the decision-making process. There was some additional evidence that decision aids resulted in fewer treatments as compared to the control group. Overall, the analysis of the large number of high-quality studies does provide evidence that decision aids can improve patient outcomes (for some specific examples for maternal patients see: Farnworth et al, 2008; Montgomery et al, 2007).
Decision Aids are Available on the Internet
There are many organizations that have implemented decision aids on the internet. Ottawa Health Decision Centre at the Ottawa Hospital Research Institute, however, brings together an extensive set of decision aids produced by a variety of organizations. This group is led by Dr. Annette O'Connor, a leading researcher in this field. Their organization's website has an "A to Z inventory" of decision aids that cover a wide variety of medical topics. To list only a few, there are extensive decision aids related to breast cancer, diabetes, headaches, acne, obesity. Patients can search an area of interest to them, and then learn more about how particular decision aids might help them with their understand their medical problem. Most of the aids posted on this site were developed by HealthWise and Health Dialog, but there are also aids developed by universities and health systems such as the Mayo Clinic. Decision aids are only posted if they meet the standards outlined by the Cochrane Collaboration. They must have development guided by expert review, have an update policy, be supported by scientific evidence, and disclose funding sources and conflicts of interest.
The Future of Internet-Based Decision Aids
There is a clear need to improve the medical decision making process, and knowledge from decision making research can be used to create evidence-based decision aids. There are, however, numerous road blocks for widespread adoption of decision aids. First, O'Connor et al. (2007) argue that many physicians do not perceive the value of decision aids. To increase adoption it will be important for medical students and practicing physicians to receive training about how decision aids might better improve their patient encounters (Holmes-Rovner et al, 2007). This will likely involve a large cultural shift in American medicine (Starr, 1982). Next, the legal system in the United States presents barriers to the use of decision aids. State laws vary however, and there are some areas that might serve as examples for states to follow. Third, payment strategies in the United States often do not provide incentives for improving patient health; capitation of payment, and possibly pay for performance strategies might encourage greater adoption of internet based decision aids. Finally, it will be important to develop and evaluate decision aids for the large number of preference-sensitive treatments in which few tools are currently available.
Although the barriers to diffusion of internet-based decision aids are real, the future for these tools seems bright. With the current debate about health reform raging, there is a slow, yet growing awareness among patients that more medicine is not always better, and physicians do not always know more than the patient. The use of the internet is also expanding, and it is encouraging, for example, that older patients at Northern California Kaiser are using online tools to manage their health (Jermy Wong, personal communication; see Cambell and Nolfi, 2005 for contrary view). Given this social environment, it is possible that patients will become more interested in using decision aids on the internet. Over the course of the next decade, they might put pressure on physicians, lawyers, and health administrations to remove barriers for use of decision aids, and enable a new generation of patient-centered care. In this new world, every healthcare organization could point their patients to decision aid web links for thousands of conditions and treatments, and counsel them to come to their appointments prepared to share with their physician personal preferences, values, and other output from these tools. Of course, with their years of training and expertise, physicians will be respected for their advice, yet they will be expected to listen carefully to what the patient has learned from the decision aid. Most importantly, physicians will be trained to understand decision science and how to help their patients value and understand decision aids. Internet-based decision aids will likely only be a small part of a reformed medical system, yet these tools will likely contribute to a health system that provides Americans equal access to a sustainable and efficient health system that does not over-treat (and harm) its people.
Blumenthal, D. (2004). Decisions, decisions: why the quality of medical decisions matters. Health Aff (Millwood), Suppl Web Exclusives, VAR124-127.
Brook, R. H., McGlynn, E. A., & Shekelle, P. G. (2000). Defining and measuring quality of care: a perspective from US researchers. Int J Qual Health Care, 12(4), 281-295.
Brownlee, S. (2007). Overtreated: why too much medicine is making us sicker and poorer (1st U.S. ed.). New York, NY: Bloomsbury.
Campbell, R. J., & Nolfi, D. A. (2005). Teaching elderly adults to use the Internet to access health care information: before-after study. J Med Internet Res, 7(2), e19.
Chassin, M. R., Kosecoff, J., Park, R. E., Winslow, C. M., Kahn, K. L., Merrick, N. J., et al. (1987). Does inappropriate use explain geographic variations in the use of health care services? A study of three procedures. Jama, 258(18), 2533-2537.
Farnworth, A., Robson, S. C., Thomson, R. G., Watson, D. B., & Murtagh, M. J. (2008). Decision support for women choosing mode of delivery after a previous caesarean section: a developmental study. Patient Educ Couns, 71(1), 116-124.
Gigerenzer, G. (2002). Calculated risks : how to know when numbers deceive you. New York: Simon & Schuster.
Holmes-Rovner, M., Nelson, W. L., Pignone, M., Elwyn, G., Rovner, D. R., O'Connor, A. M., et al. (2007). Are patient decision aids the best way to improve clinical decision making? Report of the IPDAS Symposium. Med Decis Making, 27(5), 599-608.
“How much do we know?” BMJ Clinical Evidence, 2006, http://clinicalevidence.bmj.com/ceweb/about/knowledge.jsp (accessed 16 September 2009)
Institute of Medicine (U.S.). Committee on Quality of Health Care in America. (2001). Crossing the quality chasm : a new health system for the 21st century. Washington, D.C.: National Academy Press.
Lauer, J. A., & Betran, A. P. (2007). Decision aids for women with a previous caesarean section. Bmj, 334(7607), 1281-1282.
Montgomery, A. A., Emmett, C. L., Fahey, T., Jones, C., Ricketts, I., Patel, R. R., et al. (2007). Two decision aids for mode of delivery among women with previous caesarean section: randomised controlled trial. Bmj, 334(7607), 1305.
O'Connor, A. M. (2007). Using decision aids to help patients navigate the "grey zone" of medical decision-making. CMAJ, 176(11), 1597-1598.
O'Connor, A. M., Bennett, C. L., Stacey, D., Barry, M., Col, N. F., Eden, K. B., et al. (2009). Decision aids for people facing health treatment or screening decisions. Cochrane Database Syst Rev(3), CD001431.
O'Connor, A. M., Wennberg, J. E., Legare, F., Llewellyn-Thomas, H. A., Moulton, B. W., Sepucha, K. R., et al. (2007). Toward the 'tipping point': decision aids and informed patient choice. Health Aff (Millwood), 26(3), 716-725.
Schwartz, A., & Bergus, G. (2008). Medical decision making : a physician's guide. Cambridge, UK ; New York: Cambridge University Press.
Sirovich, B., Gallagher, P. M., Wennberg, D. E., & Fisher, E. S. (2008). Discretionary decision making by primary care physicians and the cost of U.S. Health care. Health Aff (Millwood), 27(3), 813-823.
Starr, P. (1982). The social transformation of American medicine. New York: Basic Books.
Summaries for patients. Overuse of coronary artery bypass graft surgery and percutaneous transluminal coronary angioplasty. (2001). Ann Intern Med, 135(5), S-35.
Wennberg, J. E. (2004). Practice variations and health care reform: connecting the dots. Health Aff (Millwood), Suppl Web Exclusives, VAR140-144.