Using Data to Change Physician Behavior
Over the last decade, there has been an increasing focus on improving the quality of healthcare in the United States. These efforts are often aimed at utilizing data to drive physician behavior towards performance measures, evidenced-based guidelines, and quality metrics. While some quality initiatives have had some effect, gains have been modest at best and there has not been a consistently effective approach to utilizing data to drive physician behavior.1 Some of the constituent factors to this dilemma are discussed and recommendations are made to tailor these initiatives going forward towards the mindset of providers and the psychology that determines their attitudes and beliefs. Specifically these include the Theory of Planned Behavior and self-determination, but generally revolve around a better understanding of the importance of psychological ownership.
Using Data to Change Physician Behavior
With the release of several landmark healthcare quality studies at the beginning of the century2,3, there has been an ever-increasing amount of attention paid to quality improvement in healthcare. The evolution of this led to the increasing use of data to highlight the areas where quality improvement processes should be focused. After shining light on gaps in care, harm, and patient safety issues, the next logical phase was to work on finding ways to improve these areas.
In short, we started off convincing the industry that we indeed needed to improve healthcare and ever since we have been looking for ways to accomplish this endeavor. The difference between healthcare and other industries, such as manufacturing, lies in the fact that processes in healthcare are often difficult to qualify as being worthy of the distinction of the term “quality.” In other words, in healthcare the final product or outcome is not always indicative of the quality or lack of quality that led to that outcome. In fact, no matter how much we improve healthcare quality the ultimate outcome is predetermined and unavoidable. This inescapable fact leads the pursuit of improving quality to be based on concepts of outcomes such as, short-term outcomes, improved quality of life, decreased harm, etc.
Unfortunately, all of these are difficult to quantify and are built on variables that are ever-changing and based on individual beliefs and available evidence of the day. The recent efforts to digitize our electronic health records has led to an immense amount of data to be generated from these processes and does allow us to better analyze the steps that lead to these outcomes. However, when we begin to try to use this data to drive change in our clinical processes the underlying complexity in using these surrogates for quality begins to become even more evident. Even in areas where there is broad agreement over the definition of quality care, this interjects some skepticism in providers whose behavior is being analyzed. Regardless, there has been much improvement in changing the mindset of health professionals towards focusing on data and quality improvement. As we increase our efforts to improve quality by changing physician behaviors, we have to be mindful of the complexity of defining quality and incorporate this understanding and the beliefs of providers into these change management initiatives.
Data Alone is Not Truth
Data certainly is not truth, but it holds some of the greatest potential to improve healthcare quality that has ever existed. To use this most effectively we must have a better understanding of the complexities of physician behavior and how we can use data to drive this behavior toward quality care. Despite the fact that the United States healthcare system has been aggressively pursuing quality improvement for over a decade, we still have significant gaps in care even in areas where we have generally agreed upon data. Ting notes in his review article on quality improvement that the use of proven therapies in cardiovascular care varies widely from 68% to 25%.4
Further commenting on this problem, Cochrane’s review of the literature noted only a 55% adherence to evidenced based care.5 Assuming physicians are aware of these care measures, this begs the question of why then do they not do them. This is an overly simplistic evaluation of the cause and effect of non-compliance with these measures and many other factors certainly contribute to these deficits. Regardless, many of the approaches aimed at transitioning from actual care to ideal care are based upon using data to drive physician behavior. While it seems intuitive that information alone will improve behavior, this has not necessarily translated into provider behavior change. Shojania and Grimshaw highlight this in a review of the impacts of selected quality improvement strategies and found that providing information to the clinician was generally ineffective and only small benefits were seen with registries or summary of clinical performance.1 While it is not clear what causes the distinction, it does seem to hold true that interventions that incorporate interpersonal involvement, such as team changes and case management, have a greater impact than simple presentations of data, such as registries, reminders, and relays of information. 6 Data Alone is Not Enough
Recently there has been more attention paid to evaluating the failures of translating data to changes in physician behavior from a psychological or theory based approach. Francis et al, evaluated the theory of planned behavior to explain the observed effects of an intervention designed to promote evidenced-based care of diabetes. In this study, they used intention as a proxy measure of behavior and found that provider motivation translated into action based on whether or not they were attitudinally-driven or normatively-driven. The Theory of Planned Behavior model notes that these two variables along with the perception of control of the behavior will lead to the desired behavior. 7 While there are a number of reasons that determine whether or not the provider develops an attitudinally-driven or a normatively-driven mindset, the important point is to note that the desired behavior was most correlated with the development of attitudinal beliefs.7 If the provider truly believes that the intervention is important, then they are much more likely to have the desired behavior versus a provider that is motivated with the mindset that someone else is influencing them. Plainly said, this research tends to support and highlight the importance of the concept of self-determination, in that behavior is best predicted when the motivation is internal rather than external.
Similar connections to psychological ownership have been made in physician adoption of clinical information systems. Historically in clinical information system implementation, there was analysis of bad outcomes and then subsequent attempts to remedy these were made. Be it usability, support, or lack of education, there were still significant struggles even after these known areas were addressed. It was not until the importance of provider involvement in the process was stressed that true adoption or behavior change was seen consistently. In 2006, Paré correlated physician resistance to clinical information systems by exploring the psychology of how these beliefs are actually formed. By noting that psychological ownership is built from satisfying three basic motives, identified as self-enhancement, self-continuity, and a desire for a sense of control and efficacy, true adoption of the change can begin to occur.8 Whether it is the theory of self-determination or psychological ownership, the point is that there is much more to changing physician behavior than just displaying data.
Data with Incentives
Another derivation of using data for quality improvement aimed at driving physician behavior is pay for performance. Pay for performance measures are increasingly becoming a dominant component in many quality improvement initiatives and more than half of health maintenance organizations now use some form of pay for performance in the United States.9 One viewpoint of pay-for-performance incentives is that they are designed to better align the reimbursement model with the ability to provide quality care. However, many providers view this external form of motivation as a means to decrease physician reimbursement and therefore often have a negative image of these measures despite acknowledging the importance of the goals.10
In an article by Murphy et al. in the American Journal of Medical Quality, there was support for the concept of self-determination, in that they found that providers who received information on clinical performance measures and other quality initiatives from their professional specialty organization were significantly more likely to view incentives favorably. This further substantiates that if the motivation is seen as coming from within, even if it is transmitted from alignment with a professional organization, then the attitude towards the incentive is improved.10 From the previously cited articles, this attitude or stance has been associated with improved physician behavior.
An additional concern with pay for performance measures is that they could encourage gaming of the system with the adopted behavior not in line with the intent of the quality improvement initiative.9
Although there is always the possibility of this unintended consequence, it would seem to reason that providers would be much less likely to adopt this stance if they adopted ownership of the quality initiative and did not view it as much as an external pressure. This form of surrogate incentive for quality also opens the door for the physician to have to weigh the financial incentive against other possible forms of harm. Such examples could include overuse of antibiotics in conditions not requiring them, to ensure that providers do not miss any opportunities to collect reimbursement. This type of perverse incentive is one of the struggles currently faced by clinicians with concerns of patient satisfaction and defensive medicine and certainly is a consequence of pay for performance that will have to be watched closely.
Despite physicians’ innate desire to improve the lives of their patients, there are many likely reasons why simply presenting data to providers has not been broadly effective in changing behavior or increasing adoption of guidelines that are aimed at quality improvement. While much of this can be attributed to skepticism of providers surrounding area such as intent, relevance, accuracy, or other concerns about “quality” data, to truly capture the energy of providers to enact quality change, we need to adopt strategies to better leverage the accumulating mass of data with an understanding of the psychology of providers. Physicians are known for their well-developed sense of autonomy that results in a strong sense of distrust for any attempts to regulate how they care for their patients.11 Whether it is attributable to medical school training or the heuristic pattern of decision making that we learned from our mentors, it has been noted many times that “these patterns of self-directed, experiential learning…create resistance for accepting external sources of information.”12
As we move forward with our efforts to improve healthcare quality in the United States, we need to focus on this area as a determining factor of success for our initiatives. In addition to ensuring that we truly do have relevant, attributable, and accurate data, we need to anticipate practitioner concerns, provide feedback loops, and encourage and solicit input in all stages of development from design to maintenance. Ideally, we will also attempt to achieve maximum ownership of quality initiatives by allowing practitioners to pick projects that are a clinically relevant and motivating to them. Only by addressing this aspect of quality improvement will be able to truly capture the potential of data to drive physician behavior to overcome clinical quality inertia.
Tripp Jennings, MD, FACEP
- Shojania KG, Grimshaw JM. Evidence-based quality tmprovement: the state of the science. Health Affairs, Vol.24, No.1, 2005:138-150
- Committee on Quality of Health Care in America, Medicine IO. To err is human: building a safer health system. 1st Ed. National Academies Press; 2000.
- Committee on Quality of Health Care in America, Medicine IO. Crossing the quality chasm: a new health system for the 21st Century. 1st Ed. National Academies Press; 2001.
- Ting HH, Shojania KG, Montori VM, Bradley EH. Quality improvement: science and action. Circulation. 2009;119:1962–1974.
- Cochrane LJ, Olson CA, Murray S, Dupuis M, Tooman T, Hayes S. Gaps between knowing and doing: understanding and assessing the barriers to optimal health care. The Journal of Continuing Education in the Health Professions. 2007;27(2):94-102
- Shojania KG, Ranji SR, McDonald KM, Grimshaw JM, Sundaram V, Rushakoff RJ, Owens DK. Effects of quality improvement strategies for type 2 diabetes on glycemic control: a meta-regression analysis. Jama. 2006;Vol.296, No.4:427–440.
- Francis JJ, Eccles MP, Johnston M, Whitty P, Grimshaw JM, Kaner EFS, Smith L, Walker A. Explaining the effects of an intervention designed to promote evidence-based diabetes care: a theory-based process evaluation of a pragmatic cluster randomised controlled trial. Implementation Science. 2008; Vol 3. No.50.
- Paré G, Sicotte C, Jacques H. The effects of creating psychological ownership on physicians' acceptance of clinical information systems. J Am Med Inform Assoc. 2006;Vol.13.No.2:197–205.
- Mehrotra A, Damberg CL, Sorbero MES, Teleki SS. Pay for performance in the hospital setting: What is the state of the evidence? American Journal of Medical Quality. 2009 Jan;Vol 24. No.1:19-28
- Karen M. Murphy and David B. Nash. Nonprimary care physicians' views on office-based quality incentive and improvement programs. American Journal of Medical Quality 2008. Vol.23. pp.427
- Reinertsen JL, Gosfield AG, Rupp W, Whittington JW. Engaging physicians in a shared quality agenda. IHI Innovation Series white paper. Cambridge, Massachusetts: Institute for Healthcare Improvement; 2007.
- Heffner JE. Altering physician behavior to improve clinical performance. Top Health Inf Manage. 2001 Nov;Vol.22.No.2:pp.1-9.