- Top of page
- Conflicts of interest
- Source of funding
Background Medical interventions are often characterized by substantial scientific uncertainty regarding their benefits and harms. Physicians must communicate to their patients as part of the process of shared decision making, yet they may not always communicate scientific uncertainty for several reasons. One suggested by past research is individual differences in physicians’ tolerance of uncertainty. Relatedly, an unexplored explanation is physicians’ beliefs about their patients’ tolerance of uncertainty.
Design To test this possibility, we surveyed a sample of primary care physicians (N = 1500) and examined the association between their attitudes about communicating and managing scientific uncertainty and their perceptions of negative reactions to uncertainty by their patients. Physician perceptions were measured by their propensity towards pessimistic appraisals of risk information and avoidance of decision making when risk information is ambiguous – of uncertain reliability, credibility or adequacy, known as ‘ambiguity aversion’.
Results Confirming past studies, physician demographics (e.g. medical specialty) predicted attitudes toward communicating scientific uncertainty. Additionally, physicians’ beliefs about their patients’ ambiguity aversion significantly predicted these preferences. Physicians who thought that more of their patients would have negative reactions to ambiguous information were more likely to think that they should decide what is best for their patients (β = 0.065, P = 0.013), and to withhold an intervention that had uncertainty associated with it (β = 0.170, P < 0.001).
Discussion When faced with the task of communicating scientific uncertainty about medical tests and treatments, physicians’ perce-ptions of their patients’ ambiguity aversion may be related to their attitudes towards communicating uncertainty.
Scientific uncertainty is ubiquitous in health care and information about the benefits and harms of many medical interventions are thus rarely as definitive as most people would prefer. This presents a challenge for patients and physicians who must make decisions in the face of incomplete evidence and conflicting expert opinion. Ethically physicians are obligated to inform patients about uncertainty and alternative options and to encourage participation in the decision-making process at a level they desire. These obligations represent core elements of shared decision making (SDM), an accepted normative standard of health care.1–4
Despite the acknowledged importance of communicating scientific uncertainty in health care, there are significant barriers to this practice. Patients may not always prefer information and participation in decision making.5–8 Deficits in numeracy and health literacy reduce patients’ capacity to understand information and to participate effectively in decision making.9,10 Physician barriers are also important. Most physicians endorse open models of communication; however, 14% endorse limited communication of information and a paternalistic decision-making process.7 There is wide variation in the degree to which physicians actually engage in SDM under conditions of scientific uncertainty.11 Less than half of physicians’ decision-making styles appear to be concordant with patient preferences.12–14
This discrepancy between physicians’ attitudes and practices has only begun to be explained. Structural factors such as a lack of time and reimbursement have been identified as important barriers. Higher patient volume is associated with less participatory decision making.8,11,15
Individual physician characteristics also play a role. Older age has been shown to predict a lower likelihood of SDM and greater endorsement of paternalistic decision making and communication styles.16,17 Physician gender and specialty are also associated with SDM.11
Emerging evidence also suggests the influence of psychological factors. Physician attitudes regarding the value, usefulness and scientific justification of SDM vary.8,15 Physicians also differ in their capacity to tolerate or deal with uncertainty. These differences may be related to physicians’ medical training and choice of subspecialty, which is reinforced by the training and socialisation process in medical school.11,18,19 Differences in physicians’ tolerance of uncertainty are important to understand, because they may influence the extent to which physicians disclose scientific uncertainty and involve patients in decision making. Lower tolerance of uncertainty has been associated with a lower likelihood of offering a hypothetical new predictive genetic test and disclosing ambiguous test results to patients.18 This suggests that physicians’ tolerance of uncertainty may influence the extent to which they adopt strategies such as non-directiveness and information disclosure, which are more likely to respect patient autonomy.
It is not clear, however, how much these communication strategies are driven by physicians’ own tolerance of uncertainty, as opposed to perceptions of their patients’ tolerance. The perception that patients are intolerant of uncertainty might disincline physicians to fully inform patients about medical decisions marked by uncertainty.
This study examined how physicians’ perceptions of their patients’ tolerance of uncertainty, in addition to physician and practice characteristics, are related to attitudes toward disclosing information and involving patients in decisions about medical interventions characterized by scientific uncertainty. We operationalized tolerance of uncertainty in terms of ambiguity aversion (AA). ‘Ambiguity’ refers to a specific type of uncertainty pertaining to the reliability, credibility or adequacy of risk information.20,21 Ambiguity is high when information is incomplete or conflicting and leads to pessimistic appraisals of risks and choice outcomes, and avoidance of decision making.22 This response, known as ‘ambiguity aversion’, has been demonstrated in decision-making domains both in and outside of health care.23,24
We were interested in attitudes about the willingness of physicians to communicate about scientific uncertainty and to adopt a shared versus paternalistic approach towards treatment decisions under conditions of uncertainty. We predicted that physicians who perceived greater AA on the part of their patients would report less favourable attitudes towards informing patients about scientific uncertainty and involving them in decision making. We predicted a similar pattern of results related to practice experience and patient volume, such that, those with more experience would have similar attitudes to those who perceived high patient AA.11,25 We also explored the influence of other factors such as physicians’ academic affiliation and medical specialty and patients’ financial status.8,15,18,19,26,27
- Top of page
- Conflicts of interest
- Source of funding
This study examined factors associated with physicians’ attitudes towards the communication and management of scientific uncertainty in clinical practice. Physicians’ attitudes towards communicating scientific uncertainty were related not only to demographic factors, confirming past research, but also to physicians’ perceptions of their patients’ AA.
The first theme that emerged in our results related to physician demographic and practice characteristics. Professional experience was related to physicians’ attitudes towards communicating and managing scientific uncertainty. Physician’s practising longer showed less favourable attitudes towards practices consistent with open disclosure of scientific uncertainty to their patients, confirming previous findings.11 Patient volume was also related to these attitudes; physicians who saw more patients were more likely to respond that patients should have the freedom to try a new test or treatment, but also endorsed greater paternalism. This apparent contradiction might be explained pragmatically. Perhaps reflecting an important nuance in physicians’ responses to uncertainty, physicians with high patient volume may be willing to have their patients try a new intervention, but may prefer to retain final decisional authority, or to limit discussion of such topics. This is consistent with findings that that people often prefer ambiguous options if those options are perceived to be ability based, as may be the case with skilled physicians.27
Confirming past findings related to SDM practices, differences by physician specialty were also found.11,18,19 These results may reflect physicians’ experiences in treating particular patient populations and their medical training paediatricians, who in previous studies have also been found to have moderate levels of tolerance for ambiguity, showed higher endorsement of full disclosure, which may reflect their desire to divulge as much information to paediatric patients and their caregivers as possible, whereas internists showed the opposite pattern.19 Also replicating some previous findings, non-white physicians reported higher levels of paternalism and attitudes toward withholding intervention. However, other studies have found no racial differences for tolerance for ambiguity.11,19 The effect of race on the communication of uncertainty is clearly complex and must be understood in the context of its effects on physician–patient communication more broadly. For example, research on physician and patient race concordance suggests that medical interactions that are race concordant are more patient-centred.30 It is still unknown whether factors that influence the relationship between race and patient-centred communication also influence the extent of communication about scientific uncertainty.
With respect to other socio-demographic factors, an unexpected finding emerged: female physicians were more paternalistic than male physicians. Although there is a long literature indicating that men are more paternalistic in communication, a systematic review of the topic suggests that although this is usually the case, among OB/GYNs, women may actually be more paternalistic.31–34 In the present study, male physicians accounted for nearly half of the OB/GYN respondents, which may account for this seemingly contradictory finding. As many of the past studies have focused on general physician–patient communication, future research should explore gender differences in ambiguity aversion and the communication and management of scientific uncertainty.
A second theme that emerged concerned physicians’ perceptions of their patients’ responses to ambiguity. Although previous work on AA has focused on how AA can affect one’s own intentions and behaviours, ours is the first to show that perceptions of another’s AA might also have important effects. Physicians’ perceptions of their patients’ AA significantly predicted physician’s reluctance to engage in open communication about scientific uncertainty. Notably, these associations were significant over and above the effects of physician and practice characteristics. Physicians who thought more of their patients would have aversive reactions to dealing with ambiguity were less likely to communicate to their patients. This propensity is potentially important, because it would preclude giving patients the opportunity to react to uncertainty. Interestingly, physicians’ perceptions of patients’ AA were not significantly related to full disclosure of information about scientific uncertainty, although the effect was in the expected direction. However, the totality of our findings provide preliminary evidence that physicians’ perceptions of patients’ AA may influence physicians to adopt paternalistic practices and to forego communicating with patients about scientific uncertainty and involving them in decision making. This might paradoxically and counterproductively increase patients’ uncertainty and would be problematic to the extent that physicians overestimate patients’ level of ambiguity aversion.
Future research is needed to explore whether physicians’ perceptions represent accurate assessments of their patients’ attitudes, given that data are lacking on the prevalence of AA in patient populations.21 Physicians may simply project their own AA onto their patients, or rationalize their own preferences for managing scientific uncertainty. A long line of research suggests that when information about another person’s state, mood or attitude is unknown the perceiver’s perception is likely to drive behaviour.35 Other well-described psychological factors and motivations such as the power differential between physicians and patients might also contribute to physicians’ misperceptions of patients’ AA. For example, those high in power or status exhibit interpersonal perception biases; such biases could lead physicians to form inaccurate perceptions of patients’ reactions to uncertain information.36–38 However, evidence about the effect of power differentials on the accuracy of interpersonal perception is mixed.33,39 More research is necessary to determine whether cognitive biases resulting from physicians’ perceptions of their power moderate the effect of their perceptions of their patients’ AA on their attitudes.
Another question raised by our findings is the extent to which physicians’ perceptions of patients’ AA, if inaccurate, can be made either more accurate or less influential in physicians’ propensity to communicate about uncertainty. It seems desirable to improve the accuracy of physicians’ assessments of their patients’ responses to ambiguity, to promote communication strategies that are acceptable to patients and consistent with their preferences. However, a deeper question is whether patients’ AA should matter to physicians’ communication preferences. Open communication of scientific uncertainty is justified ethically by the principle of respect for patient autonomy.40 Even if disclosing information and involving patients in decision making leads to heightened risk perceptions or decisional conflict this may not be sufficient cause to forego such discussions. If, as our data suggest, physicians are truly disinclined towards such communication because they perceive their patients to be ambiguity averse, then the appropriate task at hand may not be to increase the accuracy of physicians’ perceptions. Instead, the task should be to uncouple these perceptions from physicians’ communication practices and to develop ways of improving patients’ tolerance of ambiguity. This suggests the need for meta-communication about uncertainty. That is, physicians and patients need to talk about how they talk about uncertainty. This might involve eliciting preferences for discussing such information or discussing the physician’s own tolerance for uncertainty. Such meta-communication may be a critical preliminary step towards reducing the barriers that impede SDM under the increasingly common circumstance of scientific uncertainty in health care.
Although drawn from a large nationally representative sample of physicians, conclusions are limited by the cross-sectional nature of the data, which restricts causal inferences about the relationship between physicians’ perceptions of patients’ AA and their attitudes towards SDM practices. Because the study outcomes were limited to attitudes and not actual behaviour, we do not know how our findings would translate into actual physician practice, especially given the small, although statistically significant, size of many of the effects. Although studies examining AA in patients have found consistent effects on actual behaviours, further research is needed to confirm the effects of physicians’ perceptions of patients’ AA on physician behaviours. More work is also needed to compare these effects to those of physicians’ own AA to disentangle physicians’ perceptions of their patients’ AA from their own AA.
A second limitation of this study is a potential social desirability bias of the respondents in their endorsement of the items assessing methods of communicating and managing scientific uncertainty. However, this does not appear to be a major limitation of our study, given attitudes with low social desirability, such as paternalism, received a moderate level of endorsement in the sample.
In addition, paediatricians’ mental models underlying their responses are unclear. We do not know whether paediatricians were thinking about their patients or those patients’ caregivers, who may be responsible for clinical decision making when responding to the items. The measure of ambiguity aversion also assessed physicians’ estimates of the overall proportion of their patients who were ambiguity averse; it did not ascertain physicians’ ability to discriminate between individual patients who would and would not have negative reactions to uncertainty. Nevertheless, it supports the potential influence of the physicians’ perceptions of patients’ AA on SDM practices and endorses the value of further research to elucidate this relationship.
Finally, the four main outcome variables were developed for use in the study, given the lack of validated scales to measure the extent and nature of physicians’ communication of scientific uncertainty. As designed, they exhibited low correlations with each other; however, they need to be validated in further studies. Importantly, these measures represent a novel method for assessing this process and have face validity as indicators of goals and practices fundamental to the communication of scientific uncertainty and SDM.