Strategies for Changing Clinicians’ Practice Patterns: A New Perspective

It is an all too familiar story: A review of medical records at a large primary care group practice reveals that most of the group’s physicians failed to consistently prescribe b-blockers to patients with coronary artery disease for whom they were indicated. At the next general meeting of the group’s physicians, one of the senior members makes a presentation reviewing the evidence on the indications for and benefits of prescribing b-blockers for patients with coronary artery disease. A year later a repeat review of prescribing patterns for b-blockers shows a large initial improvement for some physicians and little or no change for the rest, followed by a gradual regression toward the previous year’s overall pattern. Is this the best we can hope for?

In the past 2 decades we have learned much about changing the clinical behaviors of physicians and other clinicians. Most important, despite the field’s long-standing preference for using educational interventions to induce change in physicians’ practice patterns, we have discovered that education alone seldom leads to lasting change.^1-3 Many different change strategies have been found effective some of the time and have their adherents, but none has been found effective all the time.⁴

We still lack guidance for deciding which strategy is appropriate for a particular situation. That often leaves us at the mercy of conflicting advice.⁵ To address that gap, we present a new theoretical framework for selecting change strategies. Although we believe that several aspects of this theory are of immediate and practical use, we also recognize that our framework requires empiric testing before it can be put to full use. That work is underway in the Michigan Consortium for Family Practice Research. We present our theoretical framework here to encourage others to join us in testing and refining it.

Context and applicability

Our proposed framework addresses situations in which change is desired because one or more clinicians are not following a diagnostic or therapeutic strategy characterized by both of the following: (1) it has been proved efficacious and effective (examples include prescribing b-blockers to patients with coronary artery disease⁶ and using the Ottawa ankle rules to determine the need for radiography^7,8), and (2) there is a recurring need for it in the clinician’s practice, and any change would affect a substantial number of that clinician’s patients (eg, treating diabetic hypertension in most cardiologists’ or family physicians’ practices).

The four types of clinicians

This framework specifies that selecting the most appropriate change strategy begins with classifying each clinician according to how he or she is most likely to react to new information about the effectiveness of clinical strategies that may affect many of their patients. We distinguish 4 general categories of clinicians: the seeker, the receptive, the traditionalist, and the pragmatist.

The quintessential seeker actively reads professional journals and frequently uses electronic repositories of information. This clinician typically takes an evidence-based perspective on the literature and critically appraises papers. Seekers are as quick to abandon accepted practices when research finds them wanting as they are to adopt new ones when presented with sound evidence in their favor. They are typically not concerned about ending up out of step with locally prevalent practice patterns.Similarly, the prototypical receptive clinician is inclined to change practice in response to new information, as long as it comes from a source that indicates scientific and clinical soundness. In contrast to the seeker, who prefers to critically appraise the scientific literature independently, the receptive clinician relies on the judgments of respected authorities.

Like the receptive clinician, the typical traditionalist clinician relies on authoritative sources for guidance on whether to make changes in clinical practice in response to new information. However, because their learning style is based primarily on training and personal clinical experience, traditionalists focus on the clinical skill, experience, and authority of the advocates for change, in contrast to the receptive clinician’s greater concern with scientific arguments.

The pragmatist is a busy clinician whose concern with new information is its practicality. Any call for the pragmatist to alter some aspect of practice must be placed in the context of the many competing and often conflicting demands made by patients, colleagues, employees, insurers, and hospitals. Although pragmatists may share the seeker’s and the receptive clinician’s basic belief in being guided by what is scientifically and clinically sound or the traditionalist’s belief in clinical experience and authority, they are unlikely to willingly adopt practices if it would risk disrupting patient flow or diminishing patient satisfaction.

It is crucial to understand that these 4 categories represent traits, not states. That is, they are intended to capture characteristic patterns of response to new information. They are not depictions of moment-to-moment behavior: It is not meaningful to say that a physician was a seeker for one case and a traditionalist for another.

We have developed a short psychometric instrument for classifying physicians into our 4 categories. We tested it on 106 family physicians in Iowa and Michigan and found strong support for the theoretical construct on which the instrument is based. A revised 15-item version of the instrument will be administered in a later study to approximately 200 physicians from a range of specialties.

Categorization of change strategies

Eisenberg’s sixfold categorization¹,⁹ was among the first for viewing strategies for changing clinicians’ practices; it included education, feedback, participation, administrative changes, incentives, and penalties. More recent categorizations encompass a similar number of categories, although the labels and the focus they reflect are not necessarily the same.^4,5 We have chosen a 2-part categorization that highlights what we see as a critical distinction between commonly used strategies for affecting the knowledge of clinicians and strategies aimed at their behaviors.

Knowledge-oriented strategies are purely educational interventions epitomized by traditional continuing medical education programs. They include all other modalities for diffusing information, from hospital grand rounds to guideline dissemination.

Behavior-oriented interventions are noneducational strategies intended to alter behaviors, typified by incentives and penalties. Within this grouping we further distinguish between facilitative and directive strategies.

Facilitative behavioral strategies are used to remove barriers that stand in the way of a clinician’s adoption of new approaches to care. Too often clinicians are expected to improve care by taking actions that actually require more time and effort. Facilitative strategies can anticipate and remedy such incongruities, primarily by modifying or eliminating administrative procedures and policies likely to impede clinicians’ adherence to desired behaviors.

Directive behavioral strategies are aimed at inducing clinicians to make changes in their practices. While facilitative strategies smooth the way for change to occur, directive strategies are used to make the change actually happen. Predictably, directive strategies revolve around incentives and penalties, though not necessarily monetary ones. An example of a nonmonetary directive strategy is requiring approval from an infectious disease specialist before certain classes of antibiotics can be prescribed.

Matching change strategies to type of clinician

Our main thesis is that changing the practices of clinicians requires both knowledge-oriented and behavior-oriented strategies, but their relative roles and importance will vary for different kinds of clinicians. This is shown in the Table 1, which also illustrates how the types of interventions most commonly mentioned in the literature fit within the larger categories we specified.

Knowledge-oriented interventions are likely to be most effective in changing the practices of prototypical seekers, whose clinical behaviors can be readily influenced whenever they judge information from journal articles and presentations at professional meetings to be scientifically and clinically sound. Even though seekers are likely to make changes in spite of obstacles, virtually all are apt to respond favorably to facilitative strategies that remove obstacles not easily overcome individually, such as long-standing hospital policies or the lack of necessary equipment and other resources. Some directive strategies may also be helpful, though only marginally so, such as higher payments for adhering to evidence-based practices. Directive strategies, however, such as the promulgation of rigid rules, may be especially alienating to this kind of clinician.

Behavior-oriented strategies are the approach of choice for changing the practices of pragmatist clinicians. Chronically pressed by multiple competing demands, it is unlikely they will find time for educational sessions unless motivated by directive strategies. Educational interventions must summarize information efficiently and focus on practical issues and concerns to reach these clinicians, in contrast to the more detailed academic lectures favored by seekers. Most important, however, pragmatists will not adopt the new way of doing things if it increases their already excessive workload or conflicts with patient expectations. Hence the importance, in inducing and sustaining desired practice changes among pragmatists, of facilitative strategies to remove barriers and of directive strategies to provide appropriate incentives and rewards.

It should be noted, however, that even when the same type of intervention is expected to be equally effective for 2 types of clinicians, different strategies are likely to be needed for each. For example, the Table shows that knowledge-oriented strategies are roughly of equal importance in changing the practices of both receptive clinicians and traditionalists. Still, each type of clinician responds best to different specific interventions. The receptive clinician is most likely to heed information from scientific sources, while the traditionalist is best persuaded by information from a respected senior clinician, preferably delivered face-to-face.

Applying the framework

We believe that our perspective on changing clinical practice will be useful for guiding future efforts in the field and for research. In addition, it can help account for some of the inconsistencies and contradictions found within the literature on changing clinical practice and those found between the literature and current efforts to change clinicians’ practice patterns.

The literature on efforts to change clinical behaviors shows that almost all approaches work at least some of the time, but none works all the time.⁴ That can make this literature difficult to put to practical use by those who seek effective ways to change clinicians’ practices. We believe that our perspective could help explain what may otherwise be puzzling results. If in past studies it were possible to determine reliably the proportion of target clinicians who were seekers, receptives, traditionalists, and pragmatists, we believe most of the findings could be explained by goodness of fit or often the lack of it between the type of clinician and the characteristics of change strategies.

Our framework can similarly help explain the medical field’s enduring bias toward educational interventions, despite ample evidence that education alone seldom changes physicians’ practices.^1-4 That bias reflects an implicit assumption that all clinicians are seekers. It is an assumption readily made by the seeker-dominated groups that develop and implement change strategies. Such groups can be expected to favor the kind of purely educational interventions that our framework predicts will succeed only when directed at clinicians who also are seekers and only when there are no major practical obstacles to change. However, because most clinicians are not seekers and because there are often many obstacles to change, purely educational interventions typically fail.

In the example that began our paper, our framework predicts that the educational initiative undertaken by the senior physician would be unnecessary for the seekers in the group, who are likely to already know about b-blockers and use them as recommended. It would be effective for receptive clinicians but not for the pragmatists who may well be in the majority. Our framework predicts that even receptive clinicians will not maintain new prescribing behaviors over the long term without reinforcing facilitative measures, such as the availability of patient handouts about b-blockers and the inclusion of b-blockers on the medical group’s standard hospital discharge sheet for heart patients.

Looking ahead

The theoretical framework we have proposed includes a number of specific hypothesized relationships that are amenable to empiric testing. The psychometric instrument for assessing physician information style must be validated in a variety of practice settings, and we have begun that work. Past attempts to change physician practices, both successful and failed, need to be reviewed against the theory to determine if it accounts for their results. Prospective studies of the acceptability to physicians of various educational and behavioral interventions should show predictable variations by physician information style. Prospective trials of behavioral change interventions guided by assessment of information styles will be the ultimate test for our theory. We have begun to undertake such testing and invite and encourage others to join us in refining this framework, so that changing clinical behaviors can become a much better understood and ultimately more effective endeavor.

Acknowledgments

Based in part on work funded by grant #12549 from the Robert Wood Johnson Foundation and by the sponsorship of the Michigan Consortium for Family Practice Research by the American Academy of Family Physicians.

It is an all too familiar story: A review of medical records at a large primary care group practice reveals that most of the group’s physicians failed to consistently prescribe b-blockers to patients with coronary artery disease for whom they were indicated. At the next general meeting of the group’s physicians, one of the senior members makes a presentation reviewing the evidence on the indications for and benefits of prescribing b-blockers for patients with coronary artery disease. A year later a repeat review of prescribing patterns for b-blockers shows a large initial improvement for some physicians and little or no change for the rest, followed by a gradual regression toward the previous year’s overall pattern. Is this the best we can hope for?

In the past 2 decades we have learned much about changing the clinical behaviors of physicians and other clinicians. Most important, despite the field’s long-standing preference for using educational interventions to induce change in physicians’ practice patterns, we have discovered that education alone seldom leads to lasting change.^1-3 Many different change strategies have been found effective some of the time and have their adherents, but none has been found effective all the time.⁴

We still lack guidance for deciding which strategy is appropriate for a particular situation. That often leaves us at the mercy of conflicting advice.⁵ To address that gap, we present a new theoretical framework for selecting change strategies. Although we believe that several aspects of this theory are of immediate and practical use, we also recognize that our framework requires empiric testing before it can be put to full use. That work is underway in the Michigan Consortium for Family Practice Research. We present our theoretical framework here to encourage others to join us in testing and refining it.

Context and applicability

Our proposed framework addresses situations in which change is desired because one or more clinicians are not following a diagnostic or therapeutic strategy characterized by both of the following: (1) it has been proved efficacious and effective (examples include prescribing b-blockers to patients with coronary artery disease⁶ and using the Ottawa ankle rules to determine the need for radiography^7,8), and (2) there is a recurring need for it in the clinician’s practice, and any change would affect a substantial number of that clinician’s patients (eg, treating diabetic hypertension in most cardiologists’ or family physicians’ practices).

The four types of clinicians

This framework specifies that selecting the most appropriate change strategy begins with classifying each clinician according to how he or she is most likely to react to new information about the effectiveness of clinical strategies that may affect many of their patients. We distinguish 4 general categories of clinicians: the seeker, the receptive, the traditionalist, and the pragmatist.

The quintessential seeker actively reads professional journals and frequently uses electronic repositories of information. This clinician typically takes an evidence-based perspective on the literature and critically appraises papers. Seekers are as quick to abandon accepted practices when research finds them wanting as they are to adopt new ones when presented with sound evidence in their favor. They are typically not concerned about ending up out of step with locally prevalent practice patterns.Similarly, the prototypical receptive clinician is inclined to change practice in response to new information, as long as it comes from a source that indicates scientific and clinical soundness. In contrast to the seeker, who prefers to critically appraise the scientific literature independently, the receptive clinician relies on the judgments of respected authorities.

Like the receptive clinician, the typical traditionalist clinician relies on authoritative sources for guidance on whether to make changes in clinical practice in response to new information. However, because their learning style is based primarily on training and personal clinical experience, traditionalists focus on the clinical skill, experience, and authority of the advocates for change, in contrast to the receptive clinician’s greater concern with scientific arguments.

The pragmatist is a busy clinician whose concern with new information is its practicality. Any call for the pragmatist to alter some aspect of practice must be placed in the context of the many competing and often conflicting demands made by patients, colleagues, employees, insurers, and hospitals. Although pragmatists may share the seeker’s and the receptive clinician’s basic belief in being guided by what is scientifically and clinically sound or the traditionalist’s belief in clinical experience and authority, they are unlikely to willingly adopt practices if it would risk disrupting patient flow or diminishing patient satisfaction.

It is crucial to understand that these 4 categories represent traits, not states. That is, they are intended to capture characteristic patterns of response to new information. They are not depictions of moment-to-moment behavior: It is not meaningful to say that a physician was a seeker for one case and a traditionalist for another.

We have developed a short psychometric instrument for classifying physicians into our 4 categories. We tested it on 106 family physicians in Iowa and Michigan and found strong support for the theoretical construct on which the instrument is based. A revised 15-item version of the instrument will be administered in a later study to approximately 200 physicians from a range of specialties.

Categorization of change strategies

Eisenberg’s sixfold categorization¹,⁹ was among the first for viewing strategies for changing clinicians’ practices; it included education, feedback, participation, administrative changes, incentives, and penalties. More recent categorizations encompass a similar number of categories, although the labels and the focus they reflect are not necessarily the same.^4,5 We have chosen a 2-part categorization that highlights what we see as a critical distinction between commonly used strategies for affecting the knowledge of clinicians and strategies aimed at their behaviors.

Knowledge-oriented strategies are purely educational interventions epitomized by traditional continuing medical education programs. They include all other modalities for diffusing information, from hospital grand rounds to guideline dissemination.

Behavior-oriented interventions are noneducational strategies intended to alter behaviors, typified by incentives and penalties. Within this grouping we further distinguish between facilitative and directive strategies.

Facilitative behavioral strategies are used to remove barriers that stand in the way of a clinician’s adoption of new approaches to care. Too often clinicians are expected to improve care by taking actions that actually require more time and effort. Facilitative strategies can anticipate and remedy such incongruities, primarily by modifying or eliminating administrative procedures and policies likely to impede clinicians’ adherence to desired behaviors.

Directive behavioral strategies are aimed at inducing clinicians to make changes in their practices. While facilitative strategies smooth the way for change to occur, directive strategies are used to make the change actually happen. Predictably, directive strategies revolve around incentives and penalties, though not necessarily monetary ones. An example of a nonmonetary directive strategy is requiring approval from an infectious disease specialist before certain classes of antibiotics can be prescribed.

Matching change strategies to type of clinician

Our main thesis is that changing the practices of clinicians requires both knowledge-oriented and behavior-oriented strategies, but their relative roles and importance will vary for different kinds of clinicians. This is shown in the Table 1, which also illustrates how the types of interventions most commonly mentioned in the literature fit within the larger categories we specified.

Knowledge-oriented interventions are likely to be most effective in changing the practices of prototypical seekers, whose clinical behaviors can be readily influenced whenever they judge information from journal articles and presentations at professional meetings to be scientifically and clinically sound. Even though seekers are likely to make changes in spite of obstacles, virtually all are apt to respond favorably to facilitative strategies that remove obstacles not easily overcome individually, such as long-standing hospital policies or the lack of necessary equipment and other resources. Some directive strategies may also be helpful, though only marginally so, such as higher payments for adhering to evidence-based practices. Directive strategies, however, such as the promulgation of rigid rules, may be especially alienating to this kind of clinician.

Behavior-oriented strategies are the approach of choice for changing the practices of pragmatist clinicians. Chronically pressed by multiple competing demands, it is unlikely they will find time for educational sessions unless motivated by directive strategies. Educational interventions must summarize information efficiently and focus on practical issues and concerns to reach these clinicians, in contrast to the more detailed academic lectures favored by seekers. Most important, however, pragmatists will not adopt the new way of doing things if it increases their already excessive workload or conflicts with patient expectations. Hence the importance, in inducing and sustaining desired practice changes among pragmatists, of facilitative strategies to remove barriers and of directive strategies to provide appropriate incentives and rewards.

It should be noted, however, that even when the same type of intervention is expected to be equally effective for 2 types of clinicians, different strategies are likely to be needed for each. For example, the Table shows that knowledge-oriented strategies are roughly of equal importance in changing the practices of both receptive clinicians and traditionalists. Still, each type of clinician responds best to different specific interventions. The receptive clinician is most likely to heed information from scientific sources, while the traditionalist is best persuaded by information from a respected senior clinician, preferably delivered face-to-face.

Applying the framework

We believe that our perspective on changing clinical practice will be useful for guiding future efforts in the field and for research. In addition, it can help account for some of the inconsistencies and contradictions found within the literature on changing clinical practice and those found between the literature and current efforts to change clinicians’ practice patterns.

The literature on efforts to change clinical behaviors shows that almost all approaches work at least some of the time, but none works all the time.⁴ That can make this literature difficult to put to practical use by those who seek effective ways to change clinicians’ practices. We believe that our perspective could help explain what may otherwise be puzzling results. If in past studies it were possible to determine reliably the proportion of target clinicians who were seekers, receptives, traditionalists, and pragmatists, we believe most of the findings could be explained by goodness of fit or often the lack of it between the type of clinician and the characteristics of change strategies.

Our framework can similarly help explain the medical field’s enduring bias toward educational interventions, despite ample evidence that education alone seldom changes physicians’ practices.^1-4 That bias reflects an implicit assumption that all clinicians are seekers. It is an assumption readily made by the seeker-dominated groups that develop and implement change strategies. Such groups can be expected to favor the kind of purely educational interventions that our framework predicts will succeed only when directed at clinicians who also are seekers and only when there are no major practical obstacles to change. However, because most clinicians are not seekers and because there are often many obstacles to change, purely educational interventions typically fail.

In the example that began our paper, our framework predicts that the educational initiative undertaken by the senior physician would be unnecessary for the seekers in the group, who are likely to already know about b-blockers and use them as recommended. It would be effective for receptive clinicians but not for the pragmatists who may well be in the majority. Our framework predicts that even receptive clinicians will not maintain new prescribing behaviors over the long term without reinforcing facilitative measures, such as the availability of patient handouts about b-blockers and the inclusion of b-blockers on the medical group’s standard hospital discharge sheet for heart patients.

Looking ahead

The theoretical framework we have proposed includes a number of specific hypothesized relationships that are amenable to empiric testing. The psychometric instrument for assessing physician information style must be validated in a variety of practice settings, and we have begun that work. Past attempts to change physician practices, both successful and failed, need to be reviewed against the theory to determine if it accounts for their results. Prospective studies of the acceptability to physicians of various educational and behavioral interventions should show predictable variations by physician information style. Prospective trials of behavioral change interventions guided by assessment of information styles will be the ultimate test for our theory. We have begun to undertake such testing and invite and encourage others to join us in refining this framework, so that changing clinical behaviors can become a much better understood and ultimately more effective endeavor.

Acknowledgments

Based in part on work funded by grant #12549 from the Robert Wood Johnson Foundation and by the sponsorship of the Michigan Consortium for Family Practice Research by the American Academy of Family Physicians.

It is an all too familiar story: A review of medical records at a large primary care group practice reveals that most of the group’s physicians failed to consistently prescribe b-blockers to patients with coronary artery disease for whom they were indicated. At the next general meeting of the group’s physicians, one of the senior members makes a presentation reviewing the evidence on the indications for and benefits of prescribing b-blockers for patients with coronary artery disease. A year later a repeat review of prescribing patterns for b-blockers shows a large initial improvement for some physicians and little or no change for the rest, followed by a gradual regression toward the previous year’s overall pattern. Is this the best we can hope for?

In the past 2 decades we have learned much about changing the clinical behaviors of physicians and other clinicians. Most important, despite the field’s long-standing preference for using educational interventions to induce change in physicians’ practice patterns, we have discovered that education alone seldom leads to lasting change.^1-3 Many different change strategies have been found effective some of the time and have their adherents, but none has been found effective all the time.⁴

We still lack guidance for deciding which strategy is appropriate for a particular situation. That often leaves us at the mercy of conflicting advice.⁵ To address that gap, we present a new theoretical framework for selecting change strategies. Although we believe that several aspects of this theory are of immediate and practical use, we also recognize that our framework requires empiric testing before it can be put to full use. That work is underway in the Michigan Consortium for Family Practice Research. We present our theoretical framework here to encourage others to join us in testing and refining it.

Context and applicability

Our proposed framework addresses situations in which change is desired because one or more clinicians are not following a diagnostic or therapeutic strategy characterized by both of the following: (1) it has been proved efficacious and effective (examples include prescribing b-blockers to patients with coronary artery disease⁶ and using the Ottawa ankle rules to determine the need for radiography^7,8), and (2) there is a recurring need for it in the clinician’s practice, and any change would affect a substantial number of that clinician’s patients (eg, treating diabetic hypertension in most cardiologists’ or family physicians’ practices).

The four types of clinicians

This framework specifies that selecting the most appropriate change strategy begins with classifying each clinician according to how he or she is most likely to react to new information about the effectiveness of clinical strategies that may affect many of their patients. We distinguish 4 general categories of clinicians: the seeker, the receptive, the traditionalist, and the pragmatist.

The quintessential seeker actively reads professional journals and frequently uses electronic repositories of information. This clinician typically takes an evidence-based perspective on the literature and critically appraises papers. Seekers are as quick to abandon accepted practices when research finds them wanting as they are to adopt new ones when presented with sound evidence in their favor. They are typically not concerned about ending up out of step with locally prevalent practice patterns.Similarly, the prototypical receptive clinician is inclined to change practice in response to new information, as long as it comes from a source that indicates scientific and clinical soundness. In contrast to the seeker, who prefers to critically appraise the scientific literature independently, the receptive clinician relies on the judgments of respected authorities.

Like the receptive clinician, the typical traditionalist clinician relies on authoritative sources for guidance on whether to make changes in clinical practice in response to new information. However, because their learning style is based primarily on training and personal clinical experience, traditionalists focus on the clinical skill, experience, and authority of the advocates for change, in contrast to the receptive clinician’s greater concern with scientific arguments.

The pragmatist is a busy clinician whose concern with new information is its practicality. Any call for the pragmatist to alter some aspect of practice must be placed in the context of the many competing and often conflicting demands made by patients, colleagues, employees, insurers, and hospitals. Although pragmatists may share the seeker’s and the receptive clinician’s basic belief in being guided by what is scientifically and clinically sound or the traditionalist’s belief in clinical experience and authority, they are unlikely to willingly adopt practices if it would risk disrupting patient flow or diminishing patient satisfaction.

It is crucial to understand that these 4 categories represent traits, not states. That is, they are intended to capture characteristic patterns of response to new information. They are not depictions of moment-to-moment behavior: It is not meaningful to say that a physician was a seeker for one case and a traditionalist for another.

We have developed a short psychometric instrument for classifying physicians into our 4 categories. We tested it on 106 family physicians in Iowa and Michigan and found strong support for the theoretical construct on which the instrument is based. A revised 15-item version of the instrument will be administered in a later study to approximately 200 physicians from a range of specialties.

Categorization of change strategies

Eisenberg’s sixfold categorization¹,⁹ was among the first for viewing strategies for changing clinicians’ practices; it included education, feedback, participation, administrative changes, incentives, and penalties. More recent categorizations encompass a similar number of categories, although the labels and the focus they reflect are not necessarily the same.^4,5 We have chosen a 2-part categorization that highlights what we see as a critical distinction between commonly used strategies for affecting the knowledge of clinicians and strategies aimed at their behaviors.

Knowledge-oriented strategies are purely educational interventions epitomized by traditional continuing medical education programs. They include all other modalities for diffusing information, from hospital grand rounds to guideline dissemination.

Behavior-oriented interventions are noneducational strategies intended to alter behaviors, typified by incentives and penalties. Within this grouping we further distinguish between facilitative and directive strategies.

Facilitative behavioral strategies are used to remove barriers that stand in the way of a clinician’s adoption of new approaches to care. Too often clinicians are expected to improve care by taking actions that actually require more time and effort. Facilitative strategies can anticipate and remedy such incongruities, primarily by modifying or eliminating administrative procedures and policies likely to impede clinicians’ adherence to desired behaviors.

Directive behavioral strategies are aimed at inducing clinicians to make changes in their practices. While facilitative strategies smooth the way for change to occur, directive strategies are used to make the change actually happen. Predictably, directive strategies revolve around incentives and penalties, though not necessarily monetary ones. An example of a nonmonetary directive strategy is requiring approval from an infectious disease specialist before certain classes of antibiotics can be prescribed.

Matching change strategies to type of clinician

Our main thesis is that changing the practices of clinicians requires both knowledge-oriented and behavior-oriented strategies, but their relative roles and importance will vary for different kinds of clinicians. This is shown in the Table 1, which also illustrates how the types of interventions most commonly mentioned in the literature fit within the larger categories we specified.

Knowledge-oriented interventions are likely to be most effective in changing the practices of prototypical seekers, whose clinical behaviors can be readily influenced whenever they judge information from journal articles and presentations at professional meetings to be scientifically and clinically sound. Even though seekers are likely to make changes in spite of obstacles, virtually all are apt to respond favorably to facilitative strategies that remove obstacles not easily overcome individually, such as long-standing hospital policies or the lack of necessary equipment and other resources. Some directive strategies may also be helpful, though only marginally so, such as higher payments for adhering to evidence-based practices. Directive strategies, however, such as the promulgation of rigid rules, may be especially alienating to this kind of clinician.

Behavior-oriented strategies are the approach of choice for changing the practices of pragmatist clinicians. Chronically pressed by multiple competing demands, it is unlikely they will find time for educational sessions unless motivated by directive strategies. Educational interventions must summarize information efficiently and focus on practical issues and concerns to reach these clinicians, in contrast to the more detailed academic lectures favored by seekers. Most important, however, pragmatists will not adopt the new way of doing things if it increases their already excessive workload or conflicts with patient expectations. Hence the importance, in inducing and sustaining desired practice changes among pragmatists, of facilitative strategies to remove barriers and of directive strategies to provide appropriate incentives and rewards.

It should be noted, however, that even when the same type of intervention is expected to be equally effective for 2 types of clinicians, different strategies are likely to be needed for each. For example, the Table shows that knowledge-oriented strategies are roughly of equal importance in changing the practices of both receptive clinicians and traditionalists. Still, each type of clinician responds best to different specific interventions. The receptive clinician is most likely to heed information from scientific sources, while the traditionalist is best persuaded by information from a respected senior clinician, preferably delivered face-to-face.

Applying the framework

We believe that our perspective on changing clinical practice will be useful for guiding future efforts in the field and for research. In addition, it can help account for some of the inconsistencies and contradictions found within the literature on changing clinical practice and those found between the literature and current efforts to change clinicians’ practice patterns.

The literature on efforts to change clinical behaviors shows that almost all approaches work at least some of the time, but none works all the time.⁴ That can make this literature difficult to put to practical use by those who seek effective ways to change clinicians’ practices. We believe that our perspective could help explain what may otherwise be puzzling results. If in past studies it were possible to determine reliably the proportion of target clinicians who were seekers, receptives, traditionalists, and pragmatists, we believe most of the findings could be explained by goodness of fit or often the lack of it between the type of clinician and the characteristics of change strategies.

Our framework can similarly help explain the medical field’s enduring bias toward educational interventions, despite ample evidence that education alone seldom changes physicians’ practices.^1-4 That bias reflects an implicit assumption that all clinicians are seekers. It is an assumption readily made by the seeker-dominated groups that develop and implement change strategies. Such groups can be expected to favor the kind of purely educational interventions that our framework predicts will succeed only when directed at clinicians who also are seekers and only when there are no major practical obstacles to change. However, because most clinicians are not seekers and because there are often many obstacles to change, purely educational interventions typically fail.

In the example that began our paper, our framework predicts that the educational initiative undertaken by the senior physician would be unnecessary for the seekers in the group, who are likely to already know about b-blockers and use them as recommended. It would be effective for receptive clinicians but not for the pragmatists who may well be in the majority. Our framework predicts that even receptive clinicians will not maintain new prescribing behaviors over the long term without reinforcing facilitative measures, such as the availability of patient handouts about b-blockers and the inclusion of b-blockers on the medical group’s standard hospital discharge sheet for heart patients.

Looking ahead

The theoretical framework we have proposed includes a number of specific hypothesized relationships that are amenable to empiric testing. The psychometric instrument for assessing physician information style must be validated in a variety of practice settings, and we have begun that work. Past attempts to change physician practices, both successful and failed, need to be reviewed against the theory to determine if it accounts for their results. Prospective studies of the acceptability to physicians of various educational and behavioral interventions should show predictable variations by physician information style. Prospective trials of behavioral change interventions guided by assessment of information styles will be the ultimate test for our theory. We have begun to undertake such testing and invite and encourage others to join us in refining this framework, so that changing clinical behaviors can become a much better understood and ultimately more effective endeavor.

Acknowledgments

Based in part on work funded by grant #12549 from the Robert Wood Johnson Foundation and by the sponsorship of the Michigan Consortium for Family Practice Research by the American Academy of Family Physicians.