Perceived Barriers to Implementing Individual Choosing Wisely^® Recommendations in Two National Surveys of Primary Care Providers

What will it take to implement Choosing Wisely recommendations on a wide scale? For some recommendations, such as limiting cardiac screening before low-risk surgeries, the answer appears to be “not much,” or perhaps “we’re already doing it.”16 Very few survey respondents either perceived difficulty in following that recommendation (and the others shown in Fig. 1) or anticipated that patients would find it difficult to accept. At the same time, however, a majority of those same providers anticipated major challenges in getting patients to accept certain other CW recommendations, such as not prescribing antibiotics for sinusitis and avoiding imaging for lower back pain within the first 6 weeks. In other words, attitudes about one CW recommendation or the CW initiative as a whole cannot be assumed to predict acceptance (or lack thereof) of other recommendations.

The most conspicuous finding of our surveys was the high rates of concern about patient acceptance of the five CW recommendations shown in Figure 3, which all focus on treatment of symptomatic conditions (sinusitis, insomnia/agitation/delirium) or imaging of symptomatic conditions (syncope, low back pain). This pattern is consistent with recent research showing that hospitalists report particularly high rates of overuse of testing for syncope, and attribute this behavior to a desire to reassure patients.17

On the one hand, anticipating that patients might have concerns about recommendations related to symptomatic conditions is consistent with literature showing that patients often prioritize care for current symptomatic conditions, whereas clinicians often prioritize screening for and treatment of asymptomatic conditions.18 ^, 19 On the other hand, brief communications designed to shape patients’ mental models of their conditions might have a substantial effect on their risk perception,20 ^, 21 and hence on whether CW recommendations related to these issues are seen as acceptable. Thus, our findings are concerning, because they suggest that providers might be anticipating patient barriers to acceptance of recommendations to reduce service utilization that either do not exist or could be easily overcome. Unfortunately, such anticipatory concern could become a self-fulfilling prophecy.

We administered the survey to both a national sample of U.S. primary care physicians and a national sample of VA primary care providers in order to represent the opinions of providers practicing in the private sector and in the largest federally funded integrated health care system in the country (serving 8 million veterans). Our less than optimal response rates reflect how difficult it has become to conduct nationally representative surveys of clinicians, especially without large financial incentives.22 ^, 23 Response rate concerns also led us to conduct the surveys using different modes (mail and web), which may lead to mode effects on responses. While methodological limitations (different survey modes, timing, and sampling processes) prevent direct statistical comparison of the U.S. and VA surveys, rates of agreement in the two surveys were within a few percentage points of each other on dozens of questions, and the differences that were observed tended to involve VA physicians perceiving certain recommendations (but not others) as easier to follow, which may reflect the VA practice environment. These findings strongly support the universality of concern about implementation challenges facing efforts to reduce overuse. The findings also suggest that such concerns are not primarily driven by reimbursement issues. Yet, as with any survey, there remains the possibility that clinicians with strong opinions about overuse or the CW recommendations might disproportionately have chosen to participate or not participate in our surveys. There is also the possibility that social desirability bias induced respondents to indicate that they were more familiar with the CW initiative or more comfortable with individual recommendations than they actually were. What is clear, however, is that to whatever degree such biases occurred, it was consistent across two independent samples of clinicians operating in very different clinical environments.

The CW campaign has begun to change the discussion about how we practice medicine in the U.S. and abroad.24 Yet many providers appear to anticipate patient resistance to CW recommendations for testing or treatment of symptomatic conditions, at least in part related to patients’ requests for such services and litigation concerns. While patients may indeed find such recommendations particularly difficult to accept, it is also possible that providers’ beliefs about patient concerns may be inaccurate and that patients’ attitudes may change with effective communication. Anticipation of patient concerns should not be allowed to create undue hesitation in efforts to implement such initiatives. However, it is likely that interventions will need to extend beyond PCP-directed education, feedback, and incentives, in order to impact change for recommendations that PCPs fear patients will reject. Indeed, the cross-recommendation variations in provider attitudes that we document imply that implementation efforts will need to be tailored to the specific barriers in implementing each CW recommendation. Such tailoring may be critical in producing meaningful reductions in overuse of low-value health care services.