Decision-Aids for Prostate Cancer Screening

There is growing interest in the medical community about the development and dissemination of health care information to assist patients in making more informed choices. The proponents of this new health care consumerism assume that patients want to be informed about their choices and want to be active partners in making those decisions. Along with global advances in informatics technologies, there has been attention to and development of consumer health care informatics tools. Terms such as informed consent¹ and shared decision making² are becoming increasingly common in the medical literature and are associated with the development of many of these new consumer tools. The number of these decision-aids is increasing, as is the amount of literature evaluating their use in clinical practice.

Screening for prostate cancer has become a serious clinical concern in primary care, where physicians are the central players in an ongoing debate about offering preventive health services of unknown benefit and significant risk to patients who may not be aware of this uncertainty. Schapira and VanRuiswyk³ present a randomized comparative trial of a written decision-aid for prostate cancer screening. In their well-designed clinical trial, patients receiving an illustrated pamphlet showed greater knowledge about the accuracy of prostate cancer screening tests than did control-group patients, while no difference in the rate of screening was observed. These investigators compared 2 versions of a written pamphlet on prostate cancer screening. The comparison intervention pamphlet contained written information on prostate cancer epidemiology, symptoms, screening methods, and the benefits of screening. The decision-aid pamphlet included the same basic information plus a graphical design using human figures to represent the accuracy (sensitivity and specificity) of a combined screening strategy. The authors of this study evaluated the added impact of a graphical presentation of the accuracy of screening on patients’ knowledge, beliefs, and behaviors associated with screening for prostate cancer. It is not surprising that patients who received the decision-aid showed greater knowledge about the accuracy of prostate cancer screening.

The conceptual basis of decision-aids

How might we characterize the kind of decision-aid developed by Schapira and VanRuiswyk? In an excellent overview of the field of health care informatics and decision making, Hersey and colleagues⁴ draw a distinction between educational tools (which are preparatory and anticipatory of a decision that has already been made) and decision analysis tools (which are used to foster an informed decision by the patient). Similarly, O’Connor and coworkers⁵ use the term “tailored decision aids” to refer to patient education tools based on expected value decision theories in which models are developed to represent the structure of a decision, the probability of certain outcome events, and the patient’s valuation of those outcomes. Decision-aids can be prescriptive, using clinical decision-analysis to arrive at an optimal strategy on the basis of the expected value of the options considered. Descriptive decision-aids present probabilities and values to clarify the options and provide insight into the decision-making process.⁵ The tool developed by Schapira and VanRuiswyk would be considered a descriptive decision-aid, because it presents the probabilities of the accuracy of prostate cancer screening and encourages clarification of patients’ values associated with those outcomes.

Proponents of the paternalism model, which has dominated contemporary medicine, presume that the physician is the sole decision-maker and the patient plays a limited or no formal role in choosing a course of action. In contrast, the informed decision-making model and the shared decision-making model by Charles and colleagues⁶ include the active involvement of the patient. In the informed model, the patient is provided with all information relevant to making a decision and assumes final authority. In the shared model, patients are provided with all the pertinent information, and they work with the health care provider to come to a decision consistent with their personal values. The implementation of the decision-aid developed by Schapira and VanRuiswyk is consistent with an informed decision-making approach. Almost all decision-aids are associated with this approach.

Selecting clinical questions for decision-aids

Although it might be argued that all clinical decisions should involve various degrees of patient input, not all clinical decisions warrant the development of formal decision-aids. We suggest 3 criteria that should be met for a clinical decision to be considered appropriate for an informed decision-making intervention. First, there must be uncertainty; the optimal strategy must be unclear. Second, using a term by Kassirer,⁷ the decision must be “utility sensitive” that is, a patient’s preferences for the outcomes of treatment should be central to determining the optimal strategy. Finally, a patient’s preferences for the outcomes of treatment must vary sufficiently to warrant an individualized approach to assessment.

The question of screening for prostate cancer appears to meet these 3 criteria. There is uncertainty about the benefit of screening, and treatment holds a potential for significant complications. Previous studies using clinical decision-analysis have shown that patients’ preferences for the outcomes of prostate cancer treatment are central to determining the optimal screening strategy.^8,9 Finally, patients’ (and spouses’) preferences for these outcomes vary markedly.¹⁰

Decision-aids and prostate cancer

In the study by Schapira and VanRuiswyk, the use of prostate cancer screening after the intervention was not significantly different for the experimental and control groups (more than 80% were screened). What might explain the intervention’s lack of impact on screening behavior? A summary of clinical trials evaluating decision-aids appears to suggest that the effect of a decision-aid on screening behavior varies by subject population Table 1. In studies of unselected patients, decision-aids for prostate cancer screening appear to decrease the rate of screening. (The Mantel-Haenszel pooled relative risk estimate for these studies of unselected patients is 0.35, suggesting that decision-aids decrease screening behaviors.) Similar reductions had been observed in studies where the outcomes were intention or interest in screening.^1,11 In contrast, for studies where patients were self-referred, such as men presenting for free prostate-specific antigen testing, decision-aids appear to have little effect on screening behavior. Schapira and VanRuiswyk solicited their subjects by letter. This self-selection, as the authors note, may have led to the formation of a sample of patients who were more favorably inclined to select screening. Previous research on decision-aids suggests that a predisposition for a course of action can have an impact on the choices patients make. For example, when considering the decision to circumcise a male newborn, a decision-making tool has little effect on the rate of circumcision; parents have strong preferences before receiving the intervention and are not swayed by learning more about the risks and benefits of the procedure.⁴

Future challenges

The literature on decision-aids shows that knowledge tends to improve the situation: patients become more certain (or less conflicted) about the choices they make, and they favorably evaluate the experience.¹²

So what are the goals of informed patient decision making? O’Connor¹³ has made the astute observation that cognitively oriented decision-aids should be expected to have their greatest impact on cognitive outcomes (eg, knowledge). It seems reasonable to expect that a principal outcome of any informed decision-making intervention will be to increase patient awareness of the core issues surrounding the options they face. Reductions in decision-associated conflict, more accurate perceptions of personal risk, and satisfaction with the decision-making process are also important outcomes. Whether such interventions change behavior appears to be a secondary concern.

Perhaps the greatest challenge for this new field of patient informatics will occur as our attention turns from the efficacy of decision-aids (the effect of the intervention in highly controlled protocol-driven clinical trials) to evaluating their effectiveness (implementation in the real world of clinical practice). What seems certain is that patients will continue to want this kind of information, with many playing a more active role in decision making and looking to their health care providers for information and guidance.

There is growing interest in the medical community about the development and dissemination of health care information to assist patients in making more informed choices. The proponents of this new health care consumerism assume that patients want to be informed about their choices and want to be active partners in making those decisions. Along with global advances in informatics technologies, there has been attention to and development of consumer health care informatics tools. Terms such as informed consent¹ and shared decision making² are becoming increasingly common in the medical literature and are associated with the development of many of these new consumer tools. The number of these decision-aids is increasing, as is the amount of literature evaluating their use in clinical practice.

Screening for prostate cancer has become a serious clinical concern in primary care, where physicians are the central players in an ongoing debate about offering preventive health services of unknown benefit and significant risk to patients who may not be aware of this uncertainty. Schapira and VanRuiswyk³ present a randomized comparative trial of a written decision-aid for prostate cancer screening. In their well-designed clinical trial, patients receiving an illustrated pamphlet showed greater knowledge about the accuracy of prostate cancer screening tests than did control-group patients, while no difference in the rate of screening was observed. These investigators compared 2 versions of a written pamphlet on prostate cancer screening. The comparison intervention pamphlet contained written information on prostate cancer epidemiology, symptoms, screening methods, and the benefits of screening. The decision-aid pamphlet included the same basic information plus a graphical design using human figures to represent the accuracy (sensitivity and specificity) of a combined screening strategy. The authors of this study evaluated the added impact of a graphical presentation of the accuracy of screening on patients’ knowledge, beliefs, and behaviors associated with screening for prostate cancer. It is not surprising that patients who received the decision-aid showed greater knowledge about the accuracy of prostate cancer screening.

The conceptual basis of decision-aids

How might we characterize the kind of decision-aid developed by Schapira and VanRuiswyk? In an excellent overview of the field of health care informatics and decision making, Hersey and colleagues⁴ draw a distinction between educational tools (which are preparatory and anticipatory of a decision that has already been made) and decision analysis tools (which are used to foster an informed decision by the patient). Similarly, O’Connor and coworkers⁵ use the term “tailored decision aids” to refer to patient education tools based on expected value decision theories in which models are developed to represent the structure of a decision, the probability of certain outcome events, and the patient’s valuation of those outcomes. Decision-aids can be prescriptive, using clinical decision-analysis to arrive at an optimal strategy on the basis of the expected value of the options considered. Descriptive decision-aids present probabilities and values to clarify the options and provide insight into the decision-making process.⁵ The tool developed by Schapira and VanRuiswyk would be considered a descriptive decision-aid, because it presents the probabilities of the accuracy of prostate cancer screening and encourages clarification of patients’ values associated with those outcomes.

Proponents of the paternalism model, which has dominated contemporary medicine, presume that the physician is the sole decision-maker and the patient plays a limited or no formal role in choosing a course of action. In contrast, the informed decision-making model and the shared decision-making model by Charles and colleagues⁶ include the active involvement of the patient. In the informed model, the patient is provided with all information relevant to making a decision and assumes final authority. In the shared model, patients are provided with all the pertinent information, and they work with the health care provider to come to a decision consistent with their personal values. The implementation of the decision-aid developed by Schapira and VanRuiswyk is consistent with an informed decision-making approach. Almost all decision-aids are associated with this approach.

Selecting clinical questions for decision-aids

Although it might be argued that all clinical decisions should involve various degrees of patient input, not all clinical decisions warrant the development of formal decision-aids. We suggest 3 criteria that should be met for a clinical decision to be considered appropriate for an informed decision-making intervention. First, there must be uncertainty; the optimal strategy must be unclear. Second, using a term by Kassirer,⁷ the decision must be “utility sensitive” that is, a patient’s preferences for the outcomes of treatment should be central to determining the optimal strategy. Finally, a patient’s preferences for the outcomes of treatment must vary sufficiently to warrant an individualized approach to assessment.

The question of screening for prostate cancer appears to meet these 3 criteria. There is uncertainty about the benefit of screening, and treatment holds a potential for significant complications. Previous studies using clinical decision-analysis have shown that patients’ preferences for the outcomes of prostate cancer treatment are central to determining the optimal screening strategy.^8,9 Finally, patients’ (and spouses’) preferences for these outcomes vary markedly.¹⁰

Decision-aids and prostate cancer

In the study by Schapira and VanRuiswyk, the use of prostate cancer screening after the intervention was not significantly different for the experimental and control groups (more than 80% were screened). What might explain the intervention’s lack of impact on screening behavior? A summary of clinical trials evaluating decision-aids appears to suggest that the effect of a decision-aid on screening behavior varies by subject population Table 1. In studies of unselected patients, decision-aids for prostate cancer screening appear to decrease the rate of screening. (The Mantel-Haenszel pooled relative risk estimate for these studies of unselected patients is 0.35, suggesting that decision-aids decrease screening behaviors.) Similar reductions had been observed in studies where the outcomes were intention or interest in screening.^1,11 In contrast, for studies where patients were self-referred, such as men presenting for free prostate-specific antigen testing, decision-aids appear to have little effect on screening behavior. Schapira and VanRuiswyk solicited their subjects by letter. This self-selection, as the authors note, may have led to the formation of a sample of patients who were more favorably inclined to select screening. Previous research on decision-aids suggests that a predisposition for a course of action can have an impact on the choices patients make. For example, when considering the decision to circumcise a male newborn, a decision-making tool has little effect on the rate of circumcision; parents have strong preferences before receiving the intervention and are not swayed by learning more about the risks and benefits of the procedure.⁴

Future challenges

The literature on decision-aids shows that knowledge tends to improve the situation: patients become more certain (or less conflicted) about the choices they make, and they favorably evaluate the experience.¹²

So what are the goals of informed patient decision making? O’Connor¹³ has made the astute observation that cognitively oriented decision-aids should be expected to have their greatest impact on cognitive outcomes (eg, knowledge). It seems reasonable to expect that a principal outcome of any informed decision-making intervention will be to increase patient awareness of the core issues surrounding the options they face. Reductions in decision-associated conflict, more accurate perceptions of personal risk, and satisfaction with the decision-making process are also important outcomes. Whether such interventions change behavior appears to be a secondary concern.

Perhaps the greatest challenge for this new field of patient informatics will occur as our attention turns from the efficacy of decision-aids (the effect of the intervention in highly controlled protocol-driven clinical trials) to evaluating their effectiveness (implementation in the real world of clinical practice). What seems certain is that patients will continue to want this kind of information, with many playing a more active role in decision making and looking to their health care providers for information and guidance.

There is growing interest in the medical community about the development and dissemination of health care information to assist patients in making more informed choices. The proponents of this new health care consumerism assume that patients want to be informed about their choices and want to be active partners in making those decisions. Along with global advances in informatics technologies, there has been attention to and development of consumer health care informatics tools. Terms such as informed consent¹ and shared decision making² are becoming increasingly common in the medical literature and are associated with the development of many of these new consumer tools. The number of these decision-aids is increasing, as is the amount of literature evaluating their use in clinical practice.

Screening for prostate cancer has become a serious clinical concern in primary care, where physicians are the central players in an ongoing debate about offering preventive health services of unknown benefit and significant risk to patients who may not be aware of this uncertainty. Schapira and VanRuiswyk³ present a randomized comparative trial of a written decision-aid for prostate cancer screening. In their well-designed clinical trial, patients receiving an illustrated pamphlet showed greater knowledge about the accuracy of prostate cancer screening tests than did control-group patients, while no difference in the rate of screening was observed. These investigators compared 2 versions of a written pamphlet on prostate cancer screening. The comparison intervention pamphlet contained written information on prostate cancer epidemiology, symptoms, screening methods, and the benefits of screening. The decision-aid pamphlet included the same basic information plus a graphical design using human figures to represent the accuracy (sensitivity and specificity) of a combined screening strategy. The authors of this study evaluated the added impact of a graphical presentation of the accuracy of screening on patients’ knowledge, beliefs, and behaviors associated with screening for prostate cancer. It is not surprising that patients who received the decision-aid showed greater knowledge about the accuracy of prostate cancer screening.

The conceptual basis of decision-aids

How might we characterize the kind of decision-aid developed by Schapira and VanRuiswyk? In an excellent overview of the field of health care informatics and decision making, Hersey and colleagues⁴ draw a distinction between educational tools (which are preparatory and anticipatory of a decision that has already been made) and decision analysis tools (which are used to foster an informed decision by the patient). Similarly, O’Connor and coworkers⁵ use the term “tailored decision aids” to refer to patient education tools based on expected value decision theories in which models are developed to represent the structure of a decision, the probability of certain outcome events, and the patient’s valuation of those outcomes. Decision-aids can be prescriptive, using clinical decision-analysis to arrive at an optimal strategy on the basis of the expected value of the options considered. Descriptive decision-aids present probabilities and values to clarify the options and provide insight into the decision-making process.⁵ The tool developed by Schapira and VanRuiswyk would be considered a descriptive decision-aid, because it presents the probabilities of the accuracy of prostate cancer screening and encourages clarification of patients’ values associated with those outcomes.

Proponents of the paternalism model, which has dominated contemporary medicine, presume that the physician is the sole decision-maker and the patient plays a limited or no formal role in choosing a course of action. In contrast, the informed decision-making model and the shared decision-making model by Charles and colleagues⁶ include the active involvement of the patient. In the informed model, the patient is provided with all information relevant to making a decision and assumes final authority. In the shared model, patients are provided with all the pertinent information, and they work with the health care provider to come to a decision consistent with their personal values. The implementation of the decision-aid developed by Schapira and VanRuiswyk is consistent with an informed decision-making approach. Almost all decision-aids are associated with this approach.

Selecting clinical questions for decision-aids

Although it might be argued that all clinical decisions should involve various degrees of patient input, not all clinical decisions warrant the development of formal decision-aids. We suggest 3 criteria that should be met for a clinical decision to be considered appropriate for an informed decision-making intervention. First, there must be uncertainty; the optimal strategy must be unclear. Second, using a term by Kassirer,⁷ the decision must be “utility sensitive” that is, a patient’s preferences for the outcomes of treatment should be central to determining the optimal strategy. Finally, a patient’s preferences for the outcomes of treatment must vary sufficiently to warrant an individualized approach to assessment.

The question of screening for prostate cancer appears to meet these 3 criteria. There is uncertainty about the benefit of screening, and treatment holds a potential for significant complications. Previous studies using clinical decision-analysis have shown that patients’ preferences for the outcomes of prostate cancer treatment are central to determining the optimal screening strategy.^8,9 Finally, patients’ (and spouses’) preferences for these outcomes vary markedly.¹⁰

Decision-aids and prostate cancer

In the study by Schapira and VanRuiswyk, the use of prostate cancer screening after the intervention was not significantly different for the experimental and control groups (more than 80% were screened). What might explain the intervention’s lack of impact on screening behavior? A summary of clinical trials evaluating decision-aids appears to suggest that the effect of a decision-aid on screening behavior varies by subject population Table 1. In studies of unselected patients, decision-aids for prostate cancer screening appear to decrease the rate of screening. (The Mantel-Haenszel pooled relative risk estimate for these studies of unselected patients is 0.35, suggesting that decision-aids decrease screening behaviors.) Similar reductions had been observed in studies where the outcomes were intention or interest in screening.^1,11 In contrast, for studies where patients were self-referred, such as men presenting for free prostate-specific antigen testing, decision-aids appear to have little effect on screening behavior. Schapira and VanRuiswyk solicited their subjects by letter. This self-selection, as the authors note, may have led to the formation of a sample of patients who were more favorably inclined to select screening. Previous research on decision-aids suggests that a predisposition for a course of action can have an impact on the choices patients make. For example, when considering the decision to circumcise a male newborn, a decision-making tool has little effect on the rate of circumcision; parents have strong preferences before receiving the intervention and are not swayed by learning more about the risks and benefits of the procedure.⁴

Future challenges

The literature on decision-aids shows that knowledge tends to improve the situation: patients become more certain (or less conflicted) about the choices they make, and they favorably evaluate the experience.¹²

So what are the goals of informed patient decision making? O’Connor¹³ has made the astute observation that cognitively oriented decision-aids should be expected to have their greatest impact on cognitive outcomes (eg, knowledge). It seems reasonable to expect that a principal outcome of any informed decision-making intervention will be to increase patient awareness of the core issues surrounding the options they face. Reductions in decision-associated conflict, more accurate perceptions of personal risk, and satisfaction with the decision-making process are also important outcomes. Whether such interventions change behavior appears to be a secondary concern.

Perhaps the greatest challenge for this new field of patient informatics will occur as our attention turns from the efficacy of decision-aids (the effect of the intervention in highly controlled protocol-driven clinical trials) to evaluating their effectiveness (implementation in the real world of clinical practice). What seems certain is that patients will continue to want this kind of information, with many playing a more active role in decision making and looking to their health care providers for information and guidance.