Guidelines are developed and disseminated in the hopes that their implementation will lead to improved quality of care. However, adoption of and adherence to practice guidelines is complex [31
] and there are many steps between guideline dissemination and guideline adherence in clinical practice. Clinician awareness of a guideline is a necessary, but far from sufficient, step in this process. The ‘awareness-to-adherence’ model and similar models state that clinicians must be aware of a guideline, agree with it, adopt it as a part of care and regularly follow guideline recommendations [11
]. Cabana et al
] developed a more detailed framework—including awareness of guidelines and guideline familiarity—describing why clinicians may not adhere to clinical practice guidelines. Seemingly paradoxically, we found that self-reported familiarity with the ACP/CDC guidelines for the antibiotic treatment of ARIs was related to increased antibiotic prescribing rates, despite the fact that these guidelines discourage antibiotic use for many ARI diagnoses.
However, this apparent paradox may be the result of findings described in the psychological literature. It is generally well known that people overestimate their performance [35
]. More recently understood, to a degree that cannot be explained by regression to the mean, people who perform well on a task only slightly underestimate their performance, whereas poor performers consistently and greatly overestimate their own performance [36
]. In psychological terms, poor performers lack the metacognitive skills to know
they are underperforming; overestimating one's performance appears to be part of underperforming. Our findings may be a result of this effect: the highest antibiotic prescribers (the ‘poor performers’) were more likely to believe they were following the guidelines and, thus, overestimate their own guideline familiarity.
In the medical literature, we are not the first to note a lack of association or even an inverse relationship between self-reported familiarity or knowledge of guidelines and actual performance. Others have found that awareness of and agreement with a guideline does not guarantee knowledge of a guideline or adherence to the guideline [37
] and that clinicians are challenged to assess their own level of knowledge [38
]. Clinicians systematically overestimate their own performance compared with objective assessment by an absolute difference of about 25% [39
]. For example, consonant with the psychological literature, clinicians overestimated their adherence to hypertension guidelines, and clinicians with low guideline adherence were much more likely to overestimate their adherence to medication recommendations and blood pressure targets [40
]. With respect to ARIs, senior medical students’ compliance on hypothetical case vignettes was not correlated with their reported reading of pediatric principles of judicious antibiotic use [41
]. Our findings should be added to the literature that shows self-assessment and peer review of performance are, as compared with measuring actual performance, of questionable value [42
]. Our findings are particularly striking in that there was greater antibiotic prescribing for almost every type of ARI—both antibiotic-appropriate and antibiotic-inappropriate diagnoses—among clinicians who reported greater familiarity with the ARI guidelines.
Our study and analysis have limitations that should be considered. First, our response rate was high (65%) for a survey of mostly physicians, but not all eligible clinicians responded to the survey. Non-response could have biased the results either towards or away from the null hypothesis. Second, the survey and visit data were cross sectional. Clinicians may not have responded to the survey and treated ARI patients contemporaneously, but 85% of the survey responses we analyzed were from the first survey. In addition, about 16% of the respondents crossed over on their level of guideline familiarity from the first to the second survey. Interestingly, more respondents crossed from being more familiar to less familiar over just a few months. However, using only earlier survey results, omitting respondents who ‘crossed over,’ and adjusting for date-of-visit did not change the results. Third, we used billing codes, which are not 100% accurate for identifying ARI visits or ruling out concomitant, potentially antibiotic-appropriate diagnoses like chronic obstructive pulmonary disease. We previously found that claims-derived, electronic ARI diagnoses as a group had a sensitivity of 98%, specificity of 96% and positive predictive value of 96% [30
]. In addition, this analysis focused on the internal consistency of individual clinicians’ self-reported guideline familiarity, specified diagnoses and their actual antibiotic prescribing practices. Fourth, we are unable to control for other unmeasured confounders or factors that might be associated with self-reported guideline familiarity and antibiotic prescribing for ARIs, like overconfidence. Finally, we measured clinician self-reported familiarity with the guidelines and not actual knowledge of the guidelines. It is possible that clinicians who reported greater familiarity with the guidelines were also more knowledgeable about the guidelines and felt the guidelines did not apply to their patients.
In conclusion, we found that self-reported familiarity with the ACP/CDC guidelines for the antibiotic treatment of ARIs appeared to be associated with increased antibiotic prescribing for patients with ARIs. This should serve as an example in a new domain—outpatient management of acute medical conditions—showing that self-reported familiarity with guidelines should not be assumed to be associated with consistent guideline adherence or higher quality of care. Interventions meant to increase guideline familiarity, especially self-assessed, should not be presumed to improve the quality of care.